diff --git a/BasicPlotMaps.m b/BasicPlotMaps.m
index 9196801..f8e5c37 100644
--- a/BasicPlotMaps.m
+++ b/BasicPlotMaps.m
@@ -18,6 +18,15 @@ function BasicDrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,is
 barratio = 0.02;
 numK = double(max(double(numK),double(max(gIX))));
 
+% down-sample
+displaymax = 1000;
+numcell = size(M,1);
+if numcell > displaymax,
+    skip = round(numcell/displaymax);
+    M = M(1:skip:end,:);
+    gIX = gIX(1:skip:end,:);
+end
+
 % sort traces by index
 im = M;
 [nLines,nFrames] = size(im);
diff --git a/DrawClusters.m b/DrawClusters.m
index 964da88..524ce77 100644
--- a/DrawClusters.m
+++ b/DrawClusters.m
@@ -1,201 +1,293 @@
-function DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite)
-pos = get(gca,'Position');
-barratio = 0.03;
+function DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite,isPopout,isPlotLines,isPlotFictive)
+axis off;
+pos = get(h1,'Position'); % [left, bottom, width, height]    
 
-%% Prepare cluster data
-% down-sample
-displaymax = 1000;
-numcell = size(M,1);
-if numcell > displaymax,
-    skip = round(numcell/displaymax);
-    M = M(1:skip:end,:);
-    gIX = gIX(1:skip:end,:);
-end
-
-numK = double(max(double(numK),double(max(gIX))));
+if isPlotLines,
+    if 1, % (just to collapse isPlotLines = true)
+        %% settings
+        len = pos(3);
+        fps = 1.97;
+        
+        % set position grid
+        nLines = length(unique(gIX));
+        if isPlotFictive,
+            nExtraRows = 3; % number of extra rows
+        else
+            nExtraRows = 2;
+        end        
+        nRows = max(8,nLines)+nExtraRows;
+        lineH = pos(4)/nRows;
+%         Lpos = [pos(1),pos(2)+pos(4)-lineH*nRows,pos(3),lineH*nRows];
+% %         Lpos = [pos(1),pos(2)+pos(4)-lineH*(nLines+nExtraRows),pos(3),lineH*(nLines+nExtraRows)];
+%         set(h1,'Position',Lpos);
 
-% sort traces by index
-im = M;
-[nLines,nFrames] = size(im);
-[~,I] = sort(gIX);
-im = im(I,:);
+        %     figure('Position',[500,900-(min(nLines,15)+nDiv)*50,600,(min(nLines,15)+nDiv)*50],'color',[1 1 1]);
+        xv = 1:size(M,2);        
+                
+        %% Set colormap        
+        if strcmp(clrmap,'jet'),
+            clr = flipud(jet(double(numK)))*0.8; % make darker by *0.8
+        else % if strcmp(clrmap,'hsv'), 
+            clr = hsv(round(double(numK)*1.1))*0.7; % make darker by *0.7
+        end
+        % clr = lines(nLines);
+        % clr = ones(nLines,3)*0.3; % uniform charcoal
+        
+        %% draw stimulus bar
+        stimbar = GetStimBar(1,stim);
+        rpos = [pos(1),pos(2)+pos(4)-3/4*lineH,len,lineH/2];
+        h = axes('Position',rpos);
+        
+        image(stimbar);
+        set(h,'Xtick',[],'Ytick',[])
+        
+        % draw fish icon
+        rpos = [pos(1)*0.2,pos(2)+pos(4)-3/4*lineH,pos(1)*0.7,lineH/2];
+        axes('Position',rpos);
+        DrawFishIcon;    
+                
+        %% draw cluster means
+        Ymeans = zeros(nLines,size(M,2));
+        U = unique(gIX);
+        for j = 1:nLines,
+            k = U(j);
+            %     ymean = Cz(k,:);
+            Ys = M(gIX==k,:);
+            ymean = mean(Ys,1);
+            Ymeans(k,:) = ymean;
+            rpos = [pos(1),pos(2)+pos(4)-lineH*(j+1),len,lineH];
+            axes('Position',rpos); hold on;
+            
+            ySTD=std(Ys); % /sqrt(size(Ys,1))
+            ySTD_upper=ymean+ySTD;
+            ySTD_lower=ymean-ySTD;
+            h = fill([xv fliplr(xv)], [ySTD_upper fliplr(ySTD_lower)],0.5+0.3*clr(k,:));
+            set(h, 'EdgeColor','none')
+            
+            plot(xv,ymean,'-','Linewidth',1,'color',clr(k,:))
+            axis tight;axis off
+        end
+        
+        %% plot scale bar
+        axes('Position',[pos(1),pos(2)+pos(4)-lineH*(nLines+2),len,lineH]); hold on;
+        if size(M,2)<1200, % <~10 min, plot 1min scale bar
+            plot([1,60*fps],[0.75,0.75],'k-');
+            text(pos(1)+60*fps/2,0.5,'1 min','HorizontalAlignment','center')
+        else % >~10 min, plot 10min scale bar
+            plot([1,600*fps],[0.75,0.75],'k-');
+            text(pos(1)+600*fps/2,0.5,'10 min','HorizontalAlignment','center')
+        end
+        xlim([xv(1),xv(end)]);ylim([0,1]);
+        axis off
+                
+        %% draw fictive behavior bar
+        if isPlotFictive,
+            h = axes('Position',[pos(1),pos(2)+pos(4)-lineH*(nLines+3),len,0.7/(nLines+nExtraRows)]);
+            DrawFictiveBar(h,fictive);
+            
+            axes('Position',[0.02,pos(2)+pos(4)-lineH*(nLines+3),pos(1)-0.02,0.7/(nLines+nExtraRows)]);
+            DrawArrowsIcon(isPopout);
+        end
 
-%% convert imagesc effect into rgb matrix 'RGB'
-if dataFR, % is drawing cell-response fluorescence
-    cmap = gray(64);
-    im = AutoScaleImage0to1(im); % scaling to min/max 0/1
-    minlim = 0;
-    maxlim = 1;
-else % create blue-white-red map, for regression results
-    cmap = zeros(64,3);
-    cmap(:,1) = [linspace(0,1,32), linspace(1,1,32)];
-    cmap(:,2) = [linspace(0,1,32), linspace(1,0,32)];
-    cmap(:,3) = [linspace(1,1,32), linspace(1,0,32)];
-    minlim = -1; %min(min(im));
-    maxlim = 1; %max(max(im));
-end
-RGB = ImageToRGB(im,cmap,minlim,maxlim); % map image matrix to range of colormap
+    end
+    
+else
 
-%% add vertical color code
-if exist('clrmap','var'),
-    if strcmp(clrmap,'jet'),
-        temp = flipud(jet(numK));
+    if ~isPopout,
+        barratio = 0.025;
+    else
+        barratio = 0.05;
+    end
+    %% Prepare cluster data
+    % down-sample
+    displaymax = 1000;
+    numcell = size(M,1);
+    if numcell > displaymax,
+        skip = round(numcell/displaymax);
+        M = M(1:skip:end,:);
+        gIX = gIX(1:skip:end,:);
+    end
+    
+%     numK = double(max(double(numK),double(max(gIX))));
+    
+    % sort traces by index
+    im = M;
+    [nLines,nFrames] = size(im);
+    [~,I] = sort(gIX);
+    im = im(I,:);
+    
+    %% convert imagesc effect into rgb matrix 'RGB'
+    if dataFR, % is drawing cell-response fluorescence
+        cmap = gray(64);
+        im = AutoScaleImage0to1(im); % scaling to min/max 0/1
+        minlim = 0;
+        maxlim = 1;
+    else % create blue-white-red map, for regression results
+        cmap = zeros(64,3);
+        cmap(:,1) = [linspace(0,1,32), linspace(1,1,32)];
+        cmap(:,2) = [linspace(0,1,32), linspace(1,0,32)];
+        cmap(:,3) = [linspace(1,1,32), linspace(1,0,32)];
+        minlim = -1; %min(min(im));
+        maxlim = 1; %max(max(im));
+    end
+    RGB = ImageToRGB(im,cmap,minlim,maxlim); % map image matrix to range of colormap
+    
+    %% add vertical color code
+    if exist('clrmap','var'),
+        if strcmp(clrmap,'jet'),
+            temp = flipud(jet(numK));
+        else % 'hsv'
+            temp = hsv(round(numK*1.1));
+        end
     else % 'hsv'
         temp = hsv(round(numK*1.1));
     end
-else % 'hsv'
-    temp = hsv(round(numK*1.1));
-end
-cmap2 = vertcat(temp(1:numK,:),[0,0,0]); % extend colormap to include black
-bwidth = max(round(nFrames/30),1);
-
-idx = gIX(I);
-ix_div = [find(diff(idx));length(idx)];
-
-bars = ones(nLines,bwidth);
-if numK>1,
-    bars(1:ix_div(1),:) = idx(ix_div(1));
-    for i = 2:length(ix_div),
-        % paint color
-        bars(ix_div(i-1)+1:ix_div(i),:) = idx(ix_div(i));
-    end
-end
-im_bars = reshape(cmap2(bars,:),[size(bars),3]);
-% add a white division margin
-div = ones(nLines,round(bwidth/2),3);
-% put 3 parts together
-im = horzcat(RGB,div,im_bars);
-
-%% plot figure
-
-[s1,s2,s3] = size(im);
-
-hold off;
-set(h1,'Position',[pos(1),pos(2)+pos(4)*barratio*3,pos(3),pos(4)*(1-4*barratio)]); 
-if s1<30,
-    temp = im;
-    im = ones(30,s2,s3);
-    im(1:s1,:,:) = temp;
-end
-image(im);
-set(gca, 'box', 'off')
-
-hold on;
-
-% plot cluster division lines
-plot([0.5,s2+0.5],[0.5,0.5],'k','Linewidth',0.5);
-if numK>1,
-    for i = 1:length(ix_div),% = numK-1,
-        y = ix_div(i)+0.5;
-        plot([0.5,s2+0.5],[y,y],'k','Linewidth',0.5);
+    cmap2 = vertcat(temp(1:numK,:),[0,0,0]); % extend colormap to include black
+    bwidth = max(round(nFrames/30),1);
+    
+    idx = gIX(I);
+    ix_div = [find(diff(idx));length(idx)];
+    
+    bars = ones(nLines,bwidth);
+    if numK>1,
+        bars(1:ix_div(1),:) = idx(ix_div(1));
+        for i = 2:length(ix_div),
+            % paint color
+            bars(ix_div(i-1)+1:ix_div(i),:) = idx(ix_div(i));
+        end
     end
-end
-
-ylabel(['Number of cells: ' num2str(numcell)]);
-set(gca,'YTick',[],'XTick',[]);
-set(gcf,'color',[1 1 1]);
-
-colormap gray;
-
-% write in number label
-x = s2*1.003;
-y_last = 0;
-for i = 1:length(ix_div),
-    % avoid label crowding
-    margin = 0.015*s1;
-    y0 = ix_div(i)+0.5;
-    y = max(y_last+margin,y0);
-    if i<length(ix_div),
-        ynext = ix_div(i+1);
+    im_bars = reshape(cmap2(bars,:),[size(bars),3]);
+    % add a white division margin
+    div = ones(nLines,round(bwidth/2),3);
+    % put 3 parts together
+    im = horzcat(RGB,div,im_bars);
+    
+    %% plot figure
+    
+    [s1,s2,s3] = size(im);
+    
+    hold off;
+    if ~isPlotFictive,
+        set(h1,'Position',[pos(1),pos(2),pos(3),pos(4)*(1-barratio)]);
     else
-        ynext = y0+margin*2;
+        set(h1,'Position',[pos(1),pos(2)+pos(4)*barratio*2,pos(3),pos(4)*(1-barratio*3)]);
     end
-    % draw if not squeezed too much away
-    if y < y0+margin*2 && y < 0.5*(y0+ynext), %nLines+stimheight-margin,
-        if exist('iswrite','var') && iswrite,
-            try
-                text(x,y,[num2str(idx(ix_div(i))) ': ' num2str(rankscore(i))],'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
-                    'FontUnits','normalized','FontSize',0.015,'Tag','label');
-            catch
-                disp('rankscore invalid');
+    % axis off
+    if s1<30,
+        temp = im;
+        im = ones(30,s2,s3);
+        im(1:s1,:,:) = temp;
+    end
+    image(im);
+    set(gca, 'box', 'off')
+    
+    hold on;
+    
+    % plot cluster division lines
+    plot([0.5,s2+0.5],[0.5,0.5],'k','Linewidth',0.5);
+    if numK>1,
+        for i = 1:length(ix_div),% = numK-1,
+            y = ix_div(i)+0.5;
+            plot([0.5,s2+0.5],[y,y],'k','Linewidth',0.5);
+        end
+    end
+    
+    ylabel(['Number of cells: ' num2str(numcell)]);
+    set(gca,'YTick',[],'XTick',[]);
+    set(gcf,'color',[1 1 1]);
+    
+    colormap gray;
+    
+    % write in number label
+    x = s2*1.003;
+    y_last = 0;
+    for i = 1:length(ix_div),
+        % avoid label crowding
+        margin = 0.015*s1;
+        y0 = ix_div(i)+0.5;
+        y = max(y_last+margin,y0);
+        if i<length(ix_div),
+            ynext = ix_div(i+1);
+        else
+            ynext = y0+margin*2;
+        end
+        % draw if not squeezed too much away
+        if y < y0+margin*2 && y < 0.5*(y0+ynext), %nLines+stimheight-margin,
+            if exist('iswrite','var') && iswrite,
+                try
+                    text(x,y,[num2str(idx(ix_div(i))) ': ' num2str(rankscore(i))],'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
+                        'FontUnits','normalized','FontSize',0.015,'Tag','label');
+                catch
+                    disp('rankscore invalid');
+                    text(x,y,num2str(idx(ix_div(i))),'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
+                        'FontUnits','normalized','FontSize',0.015,'Tag','label');
+                end
+                
+            else
                 text(x,y,num2str(idx(ix_div(i))),'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
                     'FontUnits','normalized','FontSize',0.015,'Tag','label');
             end
-            
-        else
-            text(x,y,num2str(idx(ix_div(i))),'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
-                'FontUnits','normalized','FontSize',0.015,'Tag','label');
+            y_last = y;
         end
-        y_last = y;
     end
-end
-
-%% Draw stim bars
-halfbarheight = 100;
-stimbar_ = GetStimBar(halfbarheight,stim);
-
-% pad with margin on right to fit combo-image size
-barlength = size(im,2);
-stimbar = ones(halfbarheight*2,barlength,3);
-stimbar(:,1:nFrames,:) = stimbar_(:,1:nFrames,:); % crop to size
-
-% top axis
-axes('Position',[pos(1),pos(2)+pos(4)*(1-barratio),pos(3),pos(4)*barratio]);
-image(stimbar);axis off;hold on;
-plot([1,length(stim)],[1,1],'k','Linewidth',0.5); % plot top border
-plot([0.5,0.5],[0,size(stimbar,1)],'k','Linewidth',0.5); % plot left/right border
-plot([length(stim),length(stim)],[0,size(stimbar,1)],'k','Linewidth',0.5);
-
-% bottom axis
-axes('Position',[pos(1),pos(2)+pos(4)*barratio*2,pos(3),pos(4)*barratio]);
-image(stimbar);axis off;hold on;
-plot([1,length(stim)],[size(stimbar,1),size(stimbar,1)],'k','Linewidth',0.5); % plot bottom border
-plot([0.5,0.5],[0,size(stimbar,1)],'k','Linewidth',0.5); % plot left/right border
-plot([length(stim),length(stim)],[0,size(stimbar,1)],'k','Linewidth',0.5);
-
-%% Draw fictive
-axes('Position',[pos(1),pos(2),pos(3),pos(4)*barratio*2]);
-
-temp = ones(size(fictive,1),barlength);
-temp(:,1:length(fictive)) = fictive;
-imagesc(temp);colormap hot
-set(gca,'YTick',[],'XTick',[]);
-set(gcf,'color',[1 1 1]);
-set(gca, 'box', 'off')
-hold on;axis ij;
-
-% plot division lines
-for i = 0:3,
-    y = i+0.5;
-    plot([0.5,length(fictive)+0.5],[y,y],'w','Linewidth',0.5);
-end
-% labels
-names = {'Left','Right','Forward','Raw L','Raw R'};
-x = -s2*0.05;
-for i = 1:5,
-    y = i;
-    text(x,y,names{i},'Fontsize',7);
-end
-
-%% horizontal axis: approximate time length, assuming 2 fps (real is 1.97 for now)
-set(gca,'XColor',[1 1 1]);
-if s2/2/60<5, % <5 min
-    l = xlabel(['Length of display: ~ ' num2str(s2/2) 'sec'],'Color',[0 0 0]);
-else
-    l = xlabel(['Length of display: ~ ' num2str(s2/2/60) 'min'],'Color',[0 0 0]);
-end
+    
+    %% Draw stim bars
+    halfbarheight = 100;
+    stimbar_ = GetStimBar(halfbarheight,stim);
+    
+    % pad with margin on right to fit combo-image size
+    barlength = size(im,2);
+    stimbar = ones(halfbarheight*2,barlength,3);
+    stimbar(:,1:nFrames,:) = stimbar_(:,1:nFrames,:); % crop to size
+    
+    % top axis
+    
+    axes('Position',[pos(1),pos(2)+pos(4)*(1-barratio),pos(3),pos(4)*barratio]);
+    image(stimbar);axis off;hold on;
+    plot([1,length(stim)],[1,1],'k','Linewidth',0.5); % plot top border
+    plot([0.5,0.5],[0,size(stimbar,1)],'k','Linewidth',0.5); % plot left/right border
+    plot([length(stim),length(stim)],[0,size(stimbar,1)],'k','Linewidth',0.5);
+    
+    % bottom axis
+    if ~isPopout,
+        axes('Position',[pos(1),pos(2)+pos(4)*barratio*2,pos(3),pos(4)*barratio]);
+        image(stimbar);axis off;hold on;
+        plot([1,length(stim)],[size(stimbar,1),size(stimbar,1)],'k','Linewidth',0.5); % plot bottom border
+        plot([0.5,0.5],[0,size(stimbar,1)],'k','Linewidth',0.5); % plot left/right border
+        plot([length(stim),length(stim)],[0,size(stimbar,1)],'k','Linewidth',0.5);
+    end
+    %% Draw fictive
+    if isPlotFictive,
+        h = axes('Position',[pos(1),pos(2),pos(3),pos(4)*barratio*2]);        
+        DrawFictiveBar(h,fictive,barlength);        
+        axes('Position',[0.015,pos(2),0.03,pos(4)*barratio*2]);
+        DrawArrowsIcon(isPopout);
+    else
+        h = h1;
+    end
 
-%% Draw fish icon:
-if exist('fishpic.jpg','file')==2,
-    fishpic = imread('fishpic.jpg');
-    axes('Position',[0.015, 0.095, 0.03, 0.02]);
-    image(fishpic); axis off; axis equal
+    % label horizontal axis: approximate time length, assuming 1.97 fps
+    set(h,'XColor',[1 1 1]);
+    if s2/2/60<5, % <5 min
+        xlabel(['Time ~ ' num2str(round(s2/1.97)) 'sec'],'Color',[0 0 0]);
+    else
+        xlabel(['Time ~ ' num2str(round(s2/1.97/60)) 'min'],'Color',[0 0 0]);
+    end
+    
+    %% Draw fish icon:
+    if ~isPopout,
+        axes('Position',[0.015, pos(2)+pos(4)*barratio*2, 0.03, pos(4)*barratio]);
+    else
+        axes('Position',[0.015,pos(2)+pos(4)*(1-barratio),0.03,pos(4)*barratio]);
+    end
+    DrawFishIcon;
 end
 end
 
 %% Internal Functions
 
 function im = AutoScaleImage0to1(im)
+im(isnan(im)) = 0;
 temp = mat2gray(im);
 low_high = stretchlim(temp,[0.005,0.995]);
 temp(temp<low_high(1)) = low_high(1);
@@ -208,3 +300,74 @@ function DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite
 ix = round(interp1(linspace(minlim,maxlim,L),1:L,im,'linear','extrap'));
 RGB = reshape(cmap(ix,:),[size(ix) 3]); % Make RGB image from scaled.
 end
+
+function DrawFishIcon
+if exist('fishpic.jpg','file')==2,
+    pic = imread('fishpic.jpg');
+    fishpic = imresize(pic,0.5);
+    
+    image(fishpic); axis off; axis equal; axis tight
+end
+end
+
+function DrawArrowsIcon(isPopout)
+if ~isPopout,
+    if exist('arrows.jpg','file')==2,
+        pic = imread('arrows.jpg');
+        arrowpic = imresize(pic,0.2);
+        image(arrowpic); axis off; axis image; axis tight
+    end
+else
+    if exist('arrows_wide.jpg','file')==2,
+        pic = imread('arrows_wide.jpg');
+        arrowpic = imresize(pic,0.2);
+        image(arrowpic); axis off; axis image; axis tight
+    end
+end
+end
+
+function DrawFictiveBar(h,fictive,barlength)
+regressors = GetMotorRegressor(fictive);
+fc = reshape([regressors(1:5).im],length(fictive),[])';
+fc = AutoScaleImage0to1(fc);
+if exist('barlength','var'),
+    temp = ones(size(fc,1),barlength);
+    temp(:,1:length(fc)) = fc;
+else
+    temp = fc;
+end
+if 0, % plot all 5 lines
+%     imagesc(temp);colormap gray
+%     set(h,'YTick',[],'XTick',[]);
+%     %     set(gcf,'color',[1 1 1]);
+%     set(h, 'box', 'off')
+%     hold on;axis ij;
+%     
+%     % plot division lines
+%     for i = 0:3,
+%         y = i+0.5;
+%         plot([0.5,length(fictive)+0.5],[y,y],'w','Linewidth',0.5);
+%     end
+%     % labels
+%     names = {'Left','Right','Forward','Raw L','Raw R'};
+%     x = -s2*0.05;
+%     for i = 1:5,
+%         y = i;
+%         text(x,y,names{i},'Fontsize',7);
+%     end
+    
+else % only plot top 3 lines
+    fc = vertcat(temp(1,:),temp(3,:),temp(2,:));
+    imagesc(fc);colormap gray
+    set(h,'YTick',[],'XTick',[]);
+    %     set(gcf,'color',[1 1 1]);
+    set(h, 'box', 'off')
+    hold on;axis ij;
+    
+    % plot division lines
+    for i = 0:2,
+        y = i+0.5;
+        plot([0.5,length(fictive)+0.5],[y,y],'w','Linewidth',0.5);
+    end    
+end
+end
diff --git a/DrawClustersOnMap_LSh.m b/DrawClustersOnMap_LSh.m
index 5bbeb48..6ca10bf 100644
--- a/DrawClustersOnMap_LSh.m
+++ b/DrawClustersOnMap_LSh.m
@@ -27,11 +27,11 @@
 end
 
 if strcmp(clrmap,'jet'),
-    temp = flipud(jet(numK));
+    cmap = flipud(jet(numK));
 else % 'hsv'
-    temp = hsv(round(numK*1.1));
+    n = round(numK*1.1);
+    cmap = hsv(max(1,n));
 end
-cmap = temp(1:numK,:); % extend colormap to include black
 
 anat_YX = anat_yx/4;
 anat_YZ = anat_yz/4;
diff --git a/DrawTiledPics.m b/DrawTiledPics.m
new file mode 100644
index 0000000..41a22ab
--- /dev/null
+++ b/DrawTiledPics.m
@@ -0,0 +1,91 @@
+function DrawTiledPics(hfig)
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+CInfo = getappdata(hfig,'CInfo');
+ave_stack = getappdata(hfig,'ave_stack');
+
+timestamp = datestr(now,'mmddyy_HHMMSS');
+tiffName = ['stack_' timestamp '.tif'];
+
+U = unique(gIX);
+numK = length(U);
+
+% left half: stack with cells marked;
+% right half: original anatomy, or mark all cells
+
+ave_stack2=zeros(size(ave_stack,1), size(ave_stack,2), size(ave_stack,3) ,3);
+nPlanes=size(ave_stack,3);
+dimv_yxz=size(ave_stack);
+stacklen=numel(ave_stack);
+
+% circle=makeDisk2(10,21);
+radius = 10; dim = 21;
+center=floor(dim/2)+1;
+circle=zeros(dim);
+for x=1:dim
+    for y=1:dim
+        if norm([x,y]-[center,center])<=radius
+            circle(x,y)=1;
+        end
+    end
+end
+
+
+[r, v]=find(circle);
+r=r-11;v=v-11;
+circle_inds  = r*dimv_yxz(1)+v;
+cmap = hsv(round(numK*1.1));
+% cmap = [0.3 1 0];
+weight = 0.3;
+
+for i=1:nPlanes,
+    ave_stack2(:,:,i,:)=repmat(imNormalize99(ave_stack(:,:,i))/4,[1 1 1 3]);
+end
+
+for j=1:length(cIX)
+    cinds=(CInfo(cIX(j)).center(2)-1)*dimv_yxz(1)+CInfo(cIX(j)).center(1);
+    labelinds=find((cinds+circle_inds)>0 & (cinds+circle_inds)<=dimv_yxz(1)*dimv_yxz(2));
+    zinds=dimv_yxz(1)*dimv_yxz(2)*(CInfo(cIX(j)).slice-1);
+    ix = find(U==gIX(j));
+    ixs = cinds+circle_inds(labelinds)+zinds;
+    ave_stack2(ixs)=cmap(ix,1)*weight + ave_stack2(ixs)*(1-weight);
+    ixs = cinds+circle_inds(labelinds)+zinds+stacklen;
+    ave_stack2(ixs)=cmap(ix,2)*weight + ave_stack2(ixs)*(1-weight);
+    ixs = cinds+circle_inds(labelinds)+zinds+stacklen*2;
+    ave_stack2(ixs)=cmap(ix,3)*weight + ave_stack2(ixs)*(1-weight);
+end
+
+%% view stack sequentially
+% figure;
+% for i_plane = 1:nPlanes,
+%     im = squeeze(ave_stack2(:,:,i_plane,:));
+%     image(im);
+%     axis image; axis off
+%     pause(0.1)
+% end
+
+% lay out
+
+h = figure('Position',[10,20,round(1136*1.4),round(683*1.4)],'color','k');
+hold on;
+
+axes('Position',[0,0,1,1]); % BS fix to get the figure background to stay when saving pics
+image(zeros(3,3,3));axis off
+
+numRow = 3;
+numP = nPlanes - mod(nPlanes,numRow);
+numCol = numP/numRow;
+
+for i_plane = 1:numP,
+    im = squeeze(ave_stack2(:,:,i_plane,:));
+    im = imresize(im,0.25);
+    
+    [col, row] = ind2sub([numCol,numRow],i_plane); % this is intensionally inverted...
+    posVec = [(col-1)*1/numCol,1-row*1/numRow,1/numCol,1/numRow];
+    axes('Position',posVec);
+    image(im);
+    axis image; axis off
+    %     pause(0.1)
+end
+
+end
\ No newline at end of file
diff --git a/GUI_FishExplorer.m b/GUI_FishExplorer.m
index a417875..04405e0 100644
--- a/GUI_FishExplorer.m
+++ b/GUI_FishExplorer.m
@@ -21,9 +21,10 @@
 % global VAR;
 % load('VAR_current.mat','VAR');
 % load('CONSTs_current.mat','CONSTs');
-% hfig = GUI_FishExplorer(CONSTs,VAR);
+% data_dir = [pwd '\example data'];
+% [hfig,fcns] = GUI_FishExplorer(data_dir,CONSTs,VAR);
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
-% Full datasets are stored as 'CONST_F?.mat' (? = fish number), to load from the GUI
+% Full datasets are stored as 'CONST_F?_fast.mat' (? = fish number), to load from the GUI
 
 %% User Interface:
 % function hfig = GUI_FishExplorer(CONSTs,VAR,CONST,i_fish)
@@ -57,7 +58,11 @@
 %% Pass external variables into appdata (stored with main figure handle)
 setappdata(hfig,'CONSTs',CONSTs);
 setappdata(hfig,'VAR',VAR);
-nFish = length(CONSTs);
+nFish = length(CONSTs) + 1; %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+
+% fish protocol sets (different sets have different parameters)
+M_fish_set = [1, 1, 1, 1, 1, 1, 1, 2, 3, 3]; % M = Matrix
+setappdata(hfig,'M_fish_set',M_fish_set);
 
 % parameters / constants
 setappdata(hfig,'z_res',19.7); % resoltion in z, um per slice
@@ -81,9 +86,6 @@
 QuickUpdateFish(hfig,i_fish,'init');
 
 %% Initialize internal params into appdata
-% fish protocol sets (different sets have different parameters)
-M_fish_set = [1, 1, 1, 1, 1, 1, 1, 2, 1]; % M = Matrix
-setappdata(hfig,'M_fish_set',M_fish_set);
 
 % thresholds
 thres_merge = 0.9;
@@ -100,6 +102,8 @@
 setappdata(hfig,'clrmap','hsv');
 setappdata(hfig,'opID',0);
 setappdata(hfig,'rankID',0); 
+setappdata(hfig,'isPlotLines',0); 
+setappdata(hfig,'isPlotFictive',1); 
 setappdata(hfig,'rankscore',[]);
 setappdata(hfig,'isCentroid',0);
 setappdata(hfig,'isWkmeans',1); % in autoclustering, with/without kmeans
@@ -148,7 +152,7 @@
 
 % various UI element handles ('global' easier than passing around..)
 global hback hfwd hclassname hclassmenu hclusgroupmenu hclusmenu hclusname...
-    hdatamenu hopID hcentroid hdataFR hloadfish hfishnum hstimreg hmotorreg...
+    hdatamenu hopID hdataFR hloadfish hfishnum hstimreg hmotorreg...
     hcentroidreg;
 
 %% UI ----- tab one ----- (General)
@@ -176,12 +180,30 @@
     'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
     'Callback',@pushbutton_writeZstack_Callback);
 
+i=i+n;
+n=2; % plots selected cells on anatomy z-stack, tiled display
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Tile Zstack',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_tileZstack_Callback);
+
 i=i+n;
 n=2; % make new figure without the GUI components, can save manually from default menu
 uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Popup plot',...
     'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
     'Callback',@pushbutton_popupplot_Callback);
 
+i=i+n;
+n=2; % popupplot option: whether to plot cluster mean lines instead of all raw traces
+uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Plot lines',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@checkbox_isPlotLines_Callback);
+
+i=i+n;
+n=2; % popupplot option: whether to plot fictive behavior bar
+uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Plot fictive','Value',1,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@checkbox_isPlotFictive_Callback);
+
 i=i+n;
 n=3; % export main working variables to workspace, can customize!
 uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Export to workspace',...
@@ -233,7 +255,7 @@
 
 i=i+n;
 n=4; % only centroids (~mean) of clusters shown on left-side plot, the rest is unchanged
-hcentroid = uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Display centroids of clusters',...
+uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Display centroids of clusters',...
     'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
     'Callback',@checkbox_showcentroids_Callback);
 
@@ -413,7 +435,7 @@
 
 i=i+n;
 n=2; % (unlike stim regressors, names hardcoded, not importet from regressor...)
-menu = {'(choose)','right swims','left swims','forward swims','raw right','raw left','raw average'};
+menu = {'(choose)','left swims','right swims','forward swims','raw left','raw right','raw average'};
 hmotorreg = uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',menu,'Value',1,...
     'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
     'Callback',@popup_getmotorreg_Callback);
@@ -493,19 +515,19 @@
     'Callback',{@pushbutton_IterCentroidRegression_Callback});
 
 %% UI row 3: (TBD)
-i_row = 3; % old idea: display correlation values of multiple regressors (instead of fluo. trace)
-i = 1;n = 0; % and then can cluster that and further manipulate... 
-
-i=i+n;
-n=4; % did not implement combination of regressors in this version... but display option is coded (dataFR==0)
-uicontrol('Parent',tab{i_tab},'Style','text','String','regressor combos??(TBD)',... 
-    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
-
-i=i+n;
-n=4; % also data storage is coded, 'M' could be loaded from M_0_fluo or M_0_reg (storing reg corr values)
-hdataFR = uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Display fluo./regression',...
-    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
-    'Callback',@checkbox_dataFR_Callback);
+% i_row = 3; % old idea: display correlation values of multiple regressors (instead of fluo. trace)
+% i = 1;n = 0; % and then can cluster that and further manipulate... 
+% 
+% i=i+n;
+% n=4; % did not implement combination of regressors in this version... but display option is coded (dataFR==0)
+% uicontrol('Parent',tab{i_tab},'Style','text','String','regressor combos??(TBD)',... 
+%     'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+% 
+% i=i+n;
+% n=4; % also data storage is coded, 'M' could be loaded from M_0_fluo or M_0_reg (storing reg corr values)
+% hdataFR = uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Display fluo./regression',...
+%     'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+%     'Callback',@checkbox_dataFR_Callback);
 
 %% UI ----- tab four ----- (Clustering etc.)
 i_tab = 4;
@@ -977,6 +999,17 @@ function pushbutton_writeZstack_Callback(hObject,~)
 close(h)
 end
 
+function pushbutton_tileZstack_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+isfullfish = getappdata(hfig,'isfullfish');
+if isfullfish,
+    disp('Rendering...')
+    DrawTiledPics(hfig);
+else
+    errordlg('Load full fish first!');
+end
+end
+
 function out = makeDisk2(radius, dim)
 center=floor(dim/2)+1;
 out=zeros(dim);
@@ -997,7 +1030,7 @@ function pushbutton_popupplot_Callback(hObject,~)
 % figure('Position',[50,100,853,512]); % right plot size: 2048x1706
 % h1 = axes('Position',[0, 0, 0.5, 1]); % left ~subplot
 % h2 = axes('Position',[0.50, 0, 0.5, 1]); % right ~subplot
-figure('Position',[50,100,1600,900]);
+figure('Position',[50,100,1700,900],'color',[1 1 1]);
 h1 = axes('Position',[0.05, 0.03, 0.53, 0.94]); % left ~subplot
 h2 = axes('Position',[0.61, 0.03, 0.35, 0.94]); % right ~subplot
 
@@ -1017,22 +1050,39 @@ function pushbutton_popupplot_Callback(hObject,~)
 clrmap = getappdata(hfig,'clrmap');
 rankscore = getappdata(hfig,'rankscore');
 rankID = getappdata(hfig,'rankID');
-
 iswrite = (rankID>=2);
+isPlotLines = getappdata(hfig,'isPlotLines');
+isPlotFictive = getappdata(hfig,'isPlotFictive');
+
+isPopout = 1;
 
 % left subplot
 axes(h1);
 if isCentroid,
     [C,~] = FindCentroid(gIX,M);
-    DrawClusters(h1,C,unique(gIX),dataFR,numK,stim,fictive,clrmap,rankscore,iswrite);
+    DrawClusters(h1,C,unique(gIX),dataFR,numK,stim,fictive,clrmap,rankscore,...
+        iswrite,isPopout,isPlotLines,isPlotFictive);
+%     DrawClusters(h1,C,unique(gIX),dataFR,numK,stim,fictive,clrmap,rankscore,iswrite,ispopout);
 else
-    DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite);
+    DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,...
+        iswrite,isPopout,isPlotLines,isPlotFictive);
+%     DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite,ispopout);
 end
 
 % right subplot
 axes(h2);
-DrawClustersOnMap_LSh(CInfo,cIX,gIX,numK,anat_yx,anat_yz,anat_zx,clrmap);
+DrawClustersOnMap_LSh(CInfo,cIX,gIX,numK,anat_yx,anat_yz,anat_zx,clrmap,'full');
+
+end
+
+function checkbox_isPlotLines_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+setappdata(hfig,'isPlotLines',get(hObject,'Value'));
+end
 
+function checkbox_isPlotFictive_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+setappdata(hfig,'isPlotFictive',get(hObject,'Value'));
 end
 
 function pushbutton_exporttoworkspace_Callback(hObject,~)
@@ -1069,12 +1119,12 @@ function QuickUpdateFish(hfig,new_i_fish,init) %#ok<INUSD>
 numcell = CONSTs{new_i_fish}.numcell;
 
 temp = CONSTs{new_i_fish}.CRAZ;
-CRAZ = zeros(numcell,size(temp,2));
-CRAZ(CIX,:) = temp;
+CellRespAvr = zeros(numcell,size(temp,2));
+CellRespAvr(CIX,:) = temp;
 
 temp = CONSTs{new_i_fish}.CRZt;
-CRZt = zeros(numcell,size(temp,2));
-CRZt(CIX,:) = temp;
+CellResp = zeros(numcell,size(temp,2));
+CellResp(CIX,:) = temp;
 
 temp = CONSTs{new_i_fish}.CInfo;
 CInfo(numcell).center = '';
@@ -1085,17 +1135,21 @@ function QuickUpdateFish(hfig,new_i_fish,init) %#ok<INUSD>
 CInfo(numcell).x_minmax = '';
 CInfo(CIX) = temp;
 
-setappdata(hfig,'CRAZ',CRAZ); % Matrix array, misc collection
-setappdata(hfig,'CRZt',CRZt); % Matrix array, misc collection
-setappdata(hfig,'CInfo',CInfo); % Matrix array, misc collection
+setappdata(hfig,'CellRespAvr',CellRespAvr); 
+setappdata(hfig,'CellResp',CellResp); 
+setappdata(hfig,'CInfo',CInfo); 
 
 UpdateFishData(hfig,new_i_fish);
 if ~exist('init','var'),
-    UpdateFishDisplay(hfig,new_i_fish);
+    UpdateFishDisplay(hfig);
 end
 end
 
-function UpdateFishData(hfig,new_i_fish)
+function UpdateFishData(hfig,new_i_fish) % loading steps shared by both quick-load and full-load
+M_fish_set = getappdata(hfig,'M_fish_set');
+fishset = M_fish_set(new_i_fish);
+setappdata(hfig,'fishset',fishset);
+
 UpdateDatamenu_Direct(hfig,1);
 
 % save ClusGroup before updating, if applicable
@@ -1122,12 +1176,12 @@ function UpdateFishData(hfig,new_i_fish)
 setappdata(hfig,'Cluster',Cluster);
 end
 
-function UpdateFishDisplay(hfig,new_i_fish)
+function UpdateFishDisplay(hfig) % loading steps that are not performed at very first initialization
 stim = getappdata(hfig,'stim');
-M_fish_set = getappdata(hfig,'M_fish_set');
+fishset = getappdata(hfig,'fishset');
 
-fishset = M_fish_set(new_i_fish);
 [~, names] = GetStimRegressor(stim,fishset);
+
 global hstimreg;
 set(hstimreg,'String',['(choose)',(names)]);
 
@@ -1148,25 +1202,57 @@ function popup_loadfullfishmenu_Callback(hObject,~)
 set(hfishnum,'Value',new_i_fish);
 end
 
-function LoadFullFish(hfig,new_i_fish,CONST)
+function LoadFullFish(hfig,new_i_fish)
 data_dir=getappdata(hfig,'data_dir');
-if ~exist('CONST','var'),
-    disp(['loading fish #' num2str(new_i_fish) '...']);
-    tic
-    load(fullfile(data_dir,['CONST_F' num2str(new_i_fish) '.mat']),'CONST');
-    toc
-end
-setappdata(hfig,'isfullfish',1);
+disp(['loading fish #' num2str(new_i_fish) '...']);
+
+%%
+tic
+fishdir = fullfile(data_dir,['CONST_F' num2str(new_i_fish) '_fast.mat']);
+load(fishdir,'const');
+load(fishdir,'dimCR');
+CellResp = zeros(dimCR);
+num = 0;
+nParts = round(dimCR(1)*dimCR(2)/42000000);
+for i = 1:nParts,
+    load(fishdir,['CRZt_' num2str(i)']);
+    eval(['len = size(CRZt_' num2str(i) ',1);']);
+    eval(['CellResp(num+1:num+len,:) = CRZt_' num2str(i) ';']);
+    eval(['CellResp(num+1:num+len,:) = CRZt_' num2str(i) ';']);
+    num = num+len;
+end
+% for i = 1:nParts,
+%     load(fishdir,['CellResp_' num2str(i)']);
+%     eval(['len = size(CellResp_' num2str(i) ',1);']);
+%     eval(['CellResp(num+1:num+len,:) = CellResp_' num2str(i) ';']);
+%     num = num+len;
+% end
+toc
+setappdata(hfig,'CellResp',CellResp);
 
-% load all fields from CONST, with names preserved
-names = fieldnames(CONST); % cell of strings
+%% load all fields from CONST, with names preserved
+names = fieldnames(const); % cell of strings
 for i = 1:length(names),
-    setappdata(hfig,names{i},eval(['CONST.',names{i}]));
-end
-setappdata(hfig,'numcell',length(CONST.CInfo));
     
+    
+    
+    % renaming exception!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
+    if strcmp(names{i},'CRAZ'),
+        setappdata(hfig,'CellRespAvr',const.CRAZ);
+    elseif strcmp(names{i},'photostate'),
+        setappdata(hfig,'stim_full',const.photostate);
+        
+        
+        
+    else
+        setappdata(hfig,names{i},eval(['const.',names{i}]));
+    end
+end
+setappdata(hfig,'numcell',length(const.CInfo));
+
 UpdateFishData(hfig,new_i_fish);
-UpdateFishDisplay(hfig,new_i_fish);
+UpdateFishDisplay(hfig);
+setappdata(hfig,'isfullfish',1);
 end
 
 function popup_datamenu_Callback(hObject,~)
@@ -1178,29 +1264,49 @@ function popup_datamenu_Callback(hObject,~)
 end
 
 function UpdateDatamenu_Direct(hfig,ID)
-CRAZ = getappdata(hfig,'CRAZ');
-CRZt = getappdata(hfig,'CRZt');
-FcAvr = getappdata(hfig,'FcAvr');
+CellResp = getappdata(hfig,'CellResp');
 Fc = getappdata(hfig,'Fc');
 datanames = getappdata(hfig,'datanames');
 tlists = getappdata(hfig,'tlists');
-photostate = getappdata(hfig,'photostate');
+stim_full = getappdata(hfig,'stim_full');
+periods = getappdata(hfig,'periods');
+shift = getappdata(hfig,'shift');
+numcell = getappdata(hfig,'numcell');
+stimtypelist = 1:length(tlists);% getappdata(hfig,'stimtypelist');
+
+fishset = getappdata(hfig,'fishset');
+
+if fishset == 1,
+    CellRespAvr = getappdata(hfig,'CellRespAvr');
+    FcAvr = getappdata(hfig,'FcAvr');
+else
+    % find averages of different stimuli, and splice together
+    % in future should allow user to chose stimtypelist to compile average
+    % of selected stimuli groups
+    CRavr = [];
+    for i = 1:stimtypelist,
+        M = circshift(CellResp(:,tlists{i}),shift,2);
+        CRavr = horzcat(CRavr,mean(reshape(M,numcell,periods(i),[]),3));
+    end
+    setappdata(hfig,'CellRespAvr',CRavr);
+end
+
 
 %% SWITCH LEFT/RIGHT DIRECTION AGAIN
-FcAvr = vertcat(FcAvr(2,:),FcAvr(1,:),FcAvr(3:end,:));
+% % FcAvr = vertcat(FcAvr(2,:),FcAvr(1,:),FcAvr(3:end,:));
 Fc = vertcat(Fc(2,:),Fc(1,:),Fc(3:end,:));
 % now 1=left, 2=right, 3=forward, 4 = raw left, 5 = raw right
 
 %%
 if ID == 1,
-    M_0 = CRAZ;
+    M_0 = CellRespAvr;
     fictive = FcAvr;
 elseif ID == 2,
-    M_0 = CRZt;
+    M_0 = CellResp;
     fictive = Fc;
 else    
     IX = tlists{ID};
-    M_0 = CRZt(:,IX);
+    M_0 = CellResp(:,IX);
     temp = Fc(:,IX);
     % normalize fictive channels, in 2 sets
     for k = 1:3,%size(F,1),
@@ -1211,8 +1317,8 @@ function UpdateDatamenu_Direct(hfig,ID)
     temp(4:5,:) = (m-min(min(m)))/(max(max(m))-min(min(m)));
     fictive = temp;
 end
-% stim from photostate
-stim = photostate(tlists{ID}); 
+% stim from stim_full
+stim = stim_full(tlists{ID}); 
 
 setappdata(hfig,'M_0_fluo',M_0);
 setappdata(hfig,'fictive',fictive);
@@ -1411,11 +1517,12 @@ function edit_t_range_Callback(hObject,~)
 end
 end
 
-function UpdateTRange(hfig,range)
+function UpdateTRange(hfig,range) % this function is incomplete... delete?
+% what about stimulus???
+
 M_0 = GetM_0(hfig);
 cIX = getappdata(hfig,'cIX');
 fictive_0 = getappdata(hfig,'fictive_0');
-% photostate_0 = getappdata(hfig,'photostate_0');
 
 % set M
 M = M_0(cIX,range);
@@ -1423,10 +1530,6 @@ function UpdateTRange(hfig,range)
 % set fictive
 fictive = fictive_0(:,range);
 setappdata(hfig,'fictive',fictive);
-%     % set photostate
-%     photostate = photostate_0(:,range);
-%     setappdata(hfig,'photostate',photostate);
-
 end
 
 %% row 2: Operations
@@ -1504,17 +1607,17 @@ function popup_ranking_Callback(hObject,~)
 if length(periods)==1,
     period = periods{1};
     [C,~] = FindCentroid(gIX,M);
-else % i_fish>=8,
+else %if i_fish==8,
     tlists = getappdata(hfig,'tlists');
-    CRZt = getappdata(hfig,'CRZt');
+    CellResp = getappdata(hfig,'CellResp');
     IX = tlists{6}; % ptomr_circ
-    M_ = CRZt(cIX,IX);
+    M_ = CellResp(cIX,IX);
     [C,~] = FindCentroid(gIX,M_);
     periods = getappdata(hfig,'periods');
     period = periods{1}+periods{2};
+% else ?
+
 end
-% C2 = zscore(C,0,2);
-% C_3D = reshape(C2,size(C2,1),period,[]);
 C_3D_0 = reshape(C,size(C,1),period,[]);
 C_3D = zscore(C_3D_0,0,2);
 
@@ -1814,9 +1917,57 @@ function PlotRegWithStimMotor(hfig)
 reg = imNormalize99(regressor);
 regbar = repmat(reg,[1,1,3]);
 
-subplot(3,1,1);image(stimbar); axis off; title('stimulus');
-subplot(3,1,2);image(regbar); axis off; title('regressor');
-subplot(3,1,3);imagesc(fictive);colormap hot; axis off; title('motor');
+h = subplot(3,1,1);image(stimbar);set(h,'box','on','xtick',[],'ytick',[]); title('stimulus');
+h = subplot(3,1,2);image(regbar); set(h,'box','on','xtick',[],'ytick',[]); title('regressor');
+subplot(3,1,3);
+
+temp = fictive;
+if 1, % plot all 5 lines
+    fc = vertcat(temp(1,:),temp(3,:),temp(2,:),temp(4,:),temp(5,:));
+    
+    imagesc(fc);colormap gray
+    set(gca,'YTick',[],'XTick',[]);
+    set(gcf,'color',[1 1 1]);
+    set(gca, 'box', 'off')
+    hold on;axis ij;
+    
+    % plot division lines
+%     for i = 0:3,
+%         y = i+0.5;
+%         plot([0.5,length(fictive)+0.5],[y,y],'w','Linewidth',0.5);
+%     end
+%     % labels
+%     names = {'Left','Right','Forward','Raw L','Raw R'};
+%     x = -s2*0.05;
+%     for i = 1:5,
+%         y = i;
+%         text(x,y,names{i},'Fontsize',7);
+%     end
+    
+else % only plot top 3 lines
+    fc = vertcat(temp(1,:),temp(3,:),temp(2,:));
+    imagesc(fc);colormap gray
+    set(gca,'YTick',[],'XTick',[]);
+    set(gcf,'color',[1 1 1]);
+    set(gca, 'box', 'off')
+    hold on;axis ij;
+    
+    % plot division lines
+    for i = 0:2,
+        y = i+0.5;
+        plot([0.5,length(fictive)+0.5],[y,y],'w','Linewidth',0.5);
+    end
+    % labels
+    names = {'Left','Forward','Right'};
+    x = -s2*0.07;
+    for i = 1:3,
+        y = i;
+        text(x,y,names{i},'Fontsize',10);
+    end
+    
+end
+
+imagesc(fictive);colormap hot; axis off; title('motor');
 end
 
 function out=imNormalize99(im)
@@ -1881,9 +2032,7 @@ function checkbox_flipstim_Callback(hObject,~)
 stim = getappdata(hfig,'stim');
     
 if regchoice{1}==1, % stim Regressor
-    i_fish = getappdata(hfig,'i_fish');
-    M_fish_set = getappdata(hfig,'M_fish_set');
-    fishset = M_fish_set(i_fish);
+    fishset = getappdata(hfig,'fishset');
     [regressors, names] = GetStimRegressor(stim,fishset);    
     regressor = regressors(regchoice{2}).im;
     
@@ -1928,8 +2077,8 @@ function checkbox_flipstim_Callback(hObject,~)
         fliptrans = GetPhotoTrans(flipstim);
         
         % transition e.g.:
-        % photostate: 2     3     1     3     3     0     0     1     1     0     3     2     0     2     2     1
-        % phototrans: 6    11    13     7    15    12     0     1     5     4     3    14     8     2    10     9
+        % stimulus    : 2     3     1     3     3     0     0     1     1     0     3     2     0     2     2     1
+        % transitionID: 6    11    13     7    15    12     0     1     5     4     3    14     8     2    10     9
         
         IX = [];
         targetstates = unique(fliptrans,'stable');
@@ -2129,11 +2278,11 @@ function pushbutton_IterCentroidRegression_Callback(hObject,~)
 
 %% row 3: (TBD)
 
-function checkbox_dataFR_Callback(hObject,~)
-hfig = getParentFigure(hObject);
-SetDataFR(hfig,get(hObject,'Value'));
-RefreshFigure(hfig);
-end
+% function checkbox_dataFR_Callback(hObject,~) % obsolete !!!
+% hfig = getParentFigure(hObject);
+% SetDataFR(hfig,get(hObject,'Value'));
+% RefreshFigure(hfig);
+% end
 
 %% ----- tab four ----- (Clustering etc.)
 
@@ -2728,6 +2877,7 @@ function pushbutton_corrplot_Callback(hObject,~)
 M = getappdata(hfig,'M');
 [C,~] = FindCentroid(gIX,M);
 coeffs = corr(C');%corr(C(1,:)',C(2,:)')
+figure('Position',[1000,200,500,500]);
 CorrPlot(coeffs);
 end
 
@@ -2744,8 +2894,8 @@ function CorrPlot(coeffs)
 
 RGB = ImageToRGB(im,cmap,minlim,maxlim); % map image matrix to range of colormap
 
-figure('Position',[1000,200,500,500]);
 image(RGB); axis equal; axis tight;
+set(gca,'XTick',1:length(im),'YTick',1:length(im));
 
 for i = 1:size(im,2), % horizontal.. because of image axis
     for j = 1:size(im,1),
@@ -3033,7 +3183,7 @@ function SetDataFR(hfig,dataFR)
     M_0_fluo = getappdata(hfig,'M_0_fluo');
     setappdata(hfig,'M',M_0_fluo(cIX,:));
 else % Unchecked
-    M_0_reg = getappdata(hfig,'M_0_reg');
+    M_0_reg = getappdata(hfig,'M_0_reg'); % obsolete !!!
     setappdata(hfig,'M',M_0_reg(cIX,:));
 end
 global hdataFR;
@@ -3201,17 +3351,19 @@ function UpdateClusID(hfig,clusID)
             figure('Position',[100 400 1300 400]);
             subplot(1,3,1);
             CORR = corr(C');
-            imagesc(CORR);axis equal;axis tight
+            CorrPlot(CORR);
             
             subplot(1,3,2);
             dendrogram(tree,numU,'orientation','right','reorder',leafOrder);
             set(gca,'YDir','reverse');
-            colormap(jet);
+            set(gca,'XTick',[]);
+%             colormap(jet);
 
             subplot(1,3,3);
             C2 = C(leafOrder,:);
             CORR2 = corr(C2');
-            imagesc(CORR2);axis equal;axis tight
+            CorrPlot(CORR2);
+%             imagesc(CORR2);axis equal;axis tight
     end
     % sort for uniform colorscale
     temp = zeros(size(gIX));
@@ -3331,7 +3483,7 @@ function UpdateIndices(hfig,cIX,gIX,numK)
     setappdata(hfig,'cIX',cIX);
     setappdata(hfig,'gIX',gIX);
     if exist('numK','var'),
-        setappdata(hfig,'numK',numK);
+        setappdata(hfig,'numK',double(numK));
     end
 else
     errordlg('dataset too large!')
@@ -3368,8 +3520,11 @@ function RefreshFigure(hfig)
 clrmap = getappdata(hfig,'clrmap');
 rankscore = getappdata(hfig,'rankscore');
 rankID = getappdata(hfig,'rankID');
-
 iswrite = (rankID>=2);
+isPlotLines = 0; %getappdata(hfig,'isPlotLines');
+isPlotFictive = 1; %getappdata(hfig,'isPlotFictive');
+
+isPopout = 0;
 
 if isempty(cIX),
     errordlg('empty set! go back!');
@@ -3384,11 +3539,21 @@ function RefreshFigure(hfig)
 figure(hfig);
 % left subplot
 h1 = axes('Position',[0.05, 0.04, 0.55, 0.83]);
+% if isCentroid,
+%     [C,~] = FindCentroid(gIX,M);
+%     DrawClusters(h1,C,unique(gIX),dataFR,numK,stim,fictive,clrmap,rankscore,iswrite);
+% else
+%     DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite);
+% end
 if isCentroid,
     [C,~] = FindCentroid(gIX,M);
-    DrawClusters(h1,C,unique(gIX),dataFR,numK,stim,fictive,clrmap,rankscore,iswrite);
+    DrawClusters(h1,C,unique(gIX),dataFR,numK,stim,fictive,clrmap,rankscore,...
+        iswrite,isPopout,isPlotLines,isPlotFictive);
+%     DrawClusters(h1,C,unique(gIX),dataFR,numK,stim,fictive,clrmap,rankscore,iswrite,ispopout);
 else
-    DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite);
+    DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,...
+        iswrite,isPopout,isPlotLines,isPlotFictive);
+%     DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite,ispopout);
 end
 
 % right subplot
diff --git a/GUIpreload_step1_cleanup.m b/GUIpreload_step1_cleanup.m
new file mode 100644
index 0000000..f87c089
--- /dev/null
+++ b/GUIpreload_step1_cleanup.m
@@ -0,0 +1,268 @@
+%%% Loading - Step 1. 
+% Summary:
+% load data from 'cell_resp.stackf' etc, and saved into
+% 'FishX_direct_load.mat' (X = number) in the same directory.
+% Stimulus parsing is done with function 'StimulusKeyPreprocessing', but
+% one is advised to also inspect the results manually.
+
+% Validation:
+% There is an option 'isDiscard50' to discard the noisier 50%, disabled now.
+% There is also a section to manually select (draw polygons) regions of
+% cells to discard that are anatomical outliers. 
+
+% Format:
+% Cell-responses are converted into z-scores and kept that way through out.
+% CInfo (cell info for the valid cells) stores the anatomical positions.
+% For the rest of parameters saved, see the end of this file. 
+
+% Next:
+% For the following Loading steps, only this 'FishX_direct_load.mat' is 
+% needed, the rest will be cleared from working memory.
+
+%% 
+clear all;close all;clc
+
+%% Set Manaully!
+
+i_fish = 10; 
+
+M_dir = {'F:\Janelia2014\Fish1_16states_30frames';
+    'F:\Janelia2014\Fish2_20140714_2_4_16states_10frames';
+    'F:\Janelia2014\Fish3_20140715_1_1_16_states_10frames';
+    'F:\Janelia2014\Fish4_20140721_1_8_16states_20frames';
+    'F:\Janelia2014\Fish5_20140722_1_2_16states_30frames';
+    'F:\Janelia2014\Fish6_20140722_1_1_3states_30,40frames';
+    'F:\Janelia2014\Fish7_20140722_2_3_3states_30,50frames';
+    'F:\Janelia2014\Fish8_20141222_2_2_7d_PT_3OMR_shock_lowcut';
+    'F:\Janelia2014\Fish9_20150120_2_1_photo_OMR_prey_blob_blue_cy74_6d_20150120_220356';
+    'F:\Janelia2014\Fish10_20150120_2_2_photo_OMR_prey_blob_blue_cy74_6d_20150120_231917'};
+
+fpsec = 1.97; % Hz
+
+%% load data
+datadir = M_dir{i_fish};
+load(fullfile(datadir,'cell_resp_dim.mat')); % 'cell_resp_dim_lowcut.mat'
+load(fullfile(datadir,'cell_info.mat'));
+cell_resp = read_LSstack_fast_float(fullfile(datadir,'cell_resp.stackf'),cell_resp_dim);
+load(fullfile(datadir,'frame_turn.mat'));
+
+% parse stimulus - should check manually to be sure!
+[stimset,stim] = StimulusKeyPreprocessing(frame_turn);
+
+%% load anatomy
+tiffname = fullfile(datadir,'ave.tif');
+info = imfinfo(tiffname,'tiff');
+nPlanes = length(info);
+s1 = info(1).Height;
+s2 = info(1).Width;
+ave_stack = zeros(s1,s2,nPlanes);
+for i=1:nPlanes,
+    ave_stack(:,:,i) = imread(fullfile(datadir,'ave.tif'),i);
+end
+
+% x-y view
+im = max(ave_stack,[],3);
+out=imNormalize99(im);
+anat_yx = repmat(out,[1 1 3]);
+
+% y-z view
+im = squeeze(max(ave_stack,[],2));
+out=imNormalize99(im);
+anat_yz = repmat(out,[1 1 3]);
+
+% x-z view
+im = squeeze(max(ave_stack,[],1));
+out=imNormalize99(im);
+out = flipud(out'); %%%% empirically necessary...
+anat_zx = repmat(out,[1 1 3]);
+
+dimv_yx = size(anat_yx);
+dimv_yz = size(anat_yz);
+dimv_zx = size(anat_zx);
+
+%% compute z-score
+cell_resp_z = zscore(cell_resp')';
+nCells = cell_resp_dim(1);
+
+%% Validating all cells
+
+isDiscard50 = false; % option to pre-screen
+if isDiscard50,
+    %% Round 1: discard 50% noisy cells based on std of zscore of baseline
+    
+    % Compute std of dimest 10% of frames for each cell.
+    prc = prctile(cell_resp_z,10,2);
+    STD_full = zeros(nCells,1);
+    for i = 1:nCells,
+        ix = find(cell_resp_z(i,:)<prc(i));
+        STD_full(i) = std(cell_resp_z(i,ix));
+    end
+    % Set threshold at 50%, i.e. discard 50% of all cells
+    thr = prctile(STD_full,50);
+    
+    temp = find(STD_full>thr);
+    [~,I] = sort(STD_full(temp));
+    cIX = temp(I);
+    
+    % visualize cells to discard
+    % gIX = (round((1:length(cIX))/1000)+1)'; % option: view sorted index
+    gIX = round(cIX/1000)+1; % option: view anatomical z index
+    M = cell_resp_ave_z(cIX,:);
+    BasicPlotMaps(cIX,gIX,M,cell_info,stim,anat_yx,anat_yz,anat_zx);
+    
+    % I_v: index of valid cells
+    I_v_Holder = ones(1,nCells);
+    I_v_Holder(cIX) = 0;
+    I_v = find(I_v_Holder);
+    
+    CRAZ = cell_resp_ave_z(I_v,:);
+    STD = STD_full(I_v);
+    nCells = length(I_v);
+    CInfo_0 = cell_info(I_v);
+    
+    %% visualize valid cells, sorted by noise
+    [~,I] = sort(STD);
+    cIX = I;
+    gIX = (round((1:length(cIX))/1000)+1)'; % sorted index
+    M = CRAZ(I,:);
+    BasicPlotMaps(cIX,gIX,M,CInfo_0,stim,anat_yx,anat_yz,anat_zx);
+
+else    
+    %% Round 1 alternative: full set
+    I_v_Holder = ones(1,nCells);
+    I_v = (1:nCells)'; % valid Index. here not actually screened, I_v is fullset
+
+    cIX = I_v; % 'cell-Index'
+    gIX = (round((1:length(cIX))/1000)+1)'; % sorted 'group-Index'
+    M = cell_resp_z(cIX,1:1000);
+    CInfo_0 = cell_info;
+    BasicPlotMaps(cIX,gIX,M,CInfo_0,stim,anat_yx,anat_yz,anat_zx);
+end
+
+%% %%%% Round 2: delete anatomically out-of-bound cells by hand
+% only execute once (manual choice 1):
+cHolder_Anat = []; % collect cIX of all out-of-bound cells
+
+
+% draw cells first
+% (choose start and stop index to draw; low numbers ~ ventral)
+
+%% (optional) step 1: set limit to only plot ventral layer, helps to find outliers within that layer
+I_start = 1;
+I_stop = round(length(I_v)/10);
+
+% plot those cells
+cIX = I_start:I_stop;
+gIX = round(cIX/1000)+1;
+figure('Position',[100 0 1300 900]);
+numK = round(cIX(end)/1000)+1;
+[tot_image, dim_totimage] = DrawClustersOnMap_LSh(CInfo_0,cIX,gIX,numK,anat_yx,anat_yz,anat_zx,'hsv','full');
+
+% Here manually draw polygons around cells to discard. 
+% Double click to connect last vertex to first vertex, then double click again within polygon to fix.
+% then click again to start drawing the next one
+% and when finished, break loop with Ctrl+C...
+MaskArray = zeros(dim_totimage(1), dim_totimage(2));
+k_zres = 20;
+% draw polygon around outliers on Y-X view
+for i = 1:100, % NOMINAL LOOP, break manually (with Ctrl+C, somehow the design didn't work)
+    h_poly_yx = impoly;
+    wait(h_poly_yx); % double click to finalize position!
+    % update finalized polygon in bright color
+    setColor(h_poly_yx,[0 1 1]);
+
+    IJs = reshape([CInfo_0(I_start:I_stop).center],2,[])';    
+    A = sub2ind(dimv_yx(1:2),IJs(:,1),IJs(:,2));
+    MaskArray = createMask(h_poly_yx);
+    MaskArray(1:dimv_zx*k_zres+10,:) = [];
+    B = find(MaskArray); % find indices of pixels within ROI
+    cIX2 = find(ismember(A,B));
+    cHolder_Anat = union(cHolder_Anat,cIX2);
+    w = waitforbuttonpress;
+    if w == 1,
+        break; % this doesn't work ?!
+    end
+end
+
+%% step 2: plot all cells (including the ones in the ventral layer, don't 
+I_start = 1;
+I_stop = length(I_v); 
+
+% ...here until the end of the cell is the exact dupliate of the last cell...
+cIX = I_start:I_stop;
+gIX = round(cIX/1000)+1;
+figure('Position',[100 0 1300 900]);
+numK = round(cIX(end)/1000)+1;
+[tot_image, dim_totimage] = DrawClustersOnMap_LSh(CInfo_0,cIX,gIX,numK,anat_yx,anat_yz,anat_zx,'hsv','full');
+
+% Here manually draw polygons around cells to discard. 
+% Double click to connect last vertex to first vertex, then double click again within polygon to fix.
+% then click again to start drawing the next one
+% and when finished, break loop with Ctrl+C...
+MaskArray = zeros(dim_totimage(1), dim_totimage(2));
+k_zres = 20;
+% draw polygon around outliers on Y-X view
+for i = 1:100, % NOMINAL LOOP, break manually (with Ctrl+C, somehow the design didn't work)
+    h_poly_yx = impoly;
+    wait(h_poly_yx); % double click to finalize position!
+    % update finalized polygon in bright color
+    setColor(h_poly_yx,[0 1 1]);
+
+    IJs = reshape([CInfo_0(I_start:I_stop).center],2,[])';    
+    A = sub2ind(dimv_yx(1:2),IJs(:,1),IJs(:,2));
+    MaskArray = createMask(h_poly_yx);
+    MaskArray(1:dimv_zx*k_zres+10,:) = [];
+    B = find(MaskArray); % find indices of pixels within ROI
+    cIX2 = find(ismember(A,B));
+    cHolder_Anat = union(cHolder_Anat,cIX2);
+    w = waitforbuttonpress;
+    if w == 1,
+        break; % this doesn't work ?!
+    end
+end
+
+%% find extra outliers on top/bottom border of image (can't always get with polygon)
+temp = [CInfo_0(:).center];
+XY = reshape(temp',[],nCells)';
+IX = find(XY(:,1)<8 | XY(:,1)> 2040);
+
+cHolder_Anat = union(cHolder_Anat,IX);
+
+%% test Plot: all antomy outliers
+cIX = cHolder_Anat;
+gIX = (1:length(cIX))';
+figure;
+DrawClustersOnMap_LSh(CInfo_0,cIX,gIX,numK,anat_yx,anat_yz,anat_zx,'hsv','full');
+
+%% Clean up anatomy outliers
+cIX_Invalid_Anat = I_v(cHolder_Anat); % convert back to index for original full set
+I_v_Holder(cIX_Invalid_Anat) = 0;
+I_v2 = find(I_v_Holder);
+
+CR_raw = cell_resp_z(I_v2,:);
+nCells = length(I_v2);
+CInfo = cell_info(I_v2);
+
+%% Plot all valid cells to save
+cIX = (1:nCells)';
+gIX = round(cIX/1000)+1;
+M = CR_raw(cIX,1:1000);
+figure;
+DrawClustersOnMap_LSh(CInfo,cIX,gIX,numK,anat_yx,anat_yz,anat_zx,'hsv','full');
+
+%% Save mat files
+CInfo_full = cell_info; % save in next round! not saved in current mats!!!!!!
+
+temp = fullfile(datadir,['Fish' num2str(i_fish) '_full_extrainfo.mat']);
+save(temp,'cHolder_Anat','cIX_Invalid_Anat','I_v','I_v2','CInfo_full','-v7.3'); % CInfo_full not saved in mats before 4/24/15!! %% 'STD_full' saved before
+
+temp = fullfile(datadir,['Fish' num2str(i_fish) '_direct_load.mat']);
+varList = {'CR_raw','nCells','CInfo','anat_yx','anat_yz','anat_zx','ave_stack','fpsec','frame_turn'}; % used to have 'periods' handcoded
+save(temp,varList{:},'-v7.3');
+
+%% (if needed)
+% save(temp,'somethingelse','-append');
+
+%% next run step 2: detrend
+
+
diff --git a/GUIpreload_step2_detrend.m b/GUIpreload_step2_detrend.m
new file mode 100644
index 0000000..e17bdc8
--- /dev/null
+++ b/GUIpreload_step2_detrend.m
@@ -0,0 +1,66 @@
+clear all;close all;clc
+
+%% manually set directory, number of cores, range of fish (ID) to process
+
+M_dir = {'F:\Janelia2014\Fish1_16states_30frames';
+    'F:\Janelia2014\Fish2_20140714_2_4_16states_10frames';
+    'F:\Janelia2014\Fish3_20140715_1_1_16_states_10frames';
+    'F:\Janelia2014\Fish4_20140721_1_8_16states_20frames';
+    'F:\Janelia2014\Fish5_20140722_1_2_16states_30frames';
+    'F:\Janelia2014\Fish6_20140722_1_1_3states_30,40frames';
+    'F:\Janelia2014\Fish7_20140722_2_3_3states_30,50frames';
+    'F:\Janelia2014\Fish8_20141222_2_2_7d_PT_3OMR_shock_lowcut';
+    'F:\Janelia2014\Fish9_20150120_2_1_photo_OMR_prey_blob_blue_cy74_6d_20150120_220356';
+    'F:\Janelia2014\Fish10_20150120_2_2_photo_OMR_prey_blob_blue_cy74_6d_20150120_231917'};
+
+poolobj=parpool(4);
+
+range_fish = 10; %1:8
+
+%%
+for i_fish = range_fish,
+    disp(['i_fish = ', num2str(i_fish)]);
+    
+    datadir = M_dir{i_fish};
+    file = fullfile(datadir,['Fish' num2str(i_fish) '_direct_load.mat']);
+    load(file,'CR_raw');
+    %     varList = {'CR_raw','nCells','CInfo','anat_yx','anat_yz','anat_zx','ave_stack','fpsec','frame_turn','periods'};
+    
+    %% detrend
+    CR_dtr = zeros(size(CR_raw));
+    tmax=size(CR_raw,2);
+    nCells=size(CR_raw,1);
+    
+    tic
+    parfor i=1:nCells,
+        cr = CR_raw(i,:);
+        crd = 0*cr;
+        for j=1:100:tmax,
+            if j<=150,
+                tlim1 = 1;
+                tlim2 = 300;
+            elseif j>tmax-150,
+                tlim1 = tmax-300;
+                tlim2 = tmax;
+            else
+                tlim1 = j-150;
+                tlim2 = j+150;
+            end
+            crr = cr(tlim1:tlim2);
+            crd(max(1,j-50):min(tmax,j+50)) = prctile(crr,15);
+        end
+        if mod(i,100)==0,
+            disp(num2str(i));
+        end
+        CR_dtr(i,:) = zscore(cr-crd);
+    end
+    toc
+    
+    CR_dtr = single(CR_dtr);
+    
+    %% save into .mat
+    save(file,'CR_dtr','-append');
+end
+
+%%
+delete(poolobj);
diff --git a/GUIpreload_step3.5_PoolClustersFromAllFish.m b/GUIpreload_step3.5_PoolClustersFromAllFish.m
new file mode 100644
index 0000000..6ddcbc9
--- /dev/null
+++ b/GUIpreload_step3.5_PoolClustersFromAllFish.m
@@ -0,0 +1,113 @@
+% Pool from all fish
+clear all;close all;clc
+
+data_dir = 'F:\Janelia2014';
+
+global VAR;
+load('VAR_current.mat','VAR');
+%%
+numfish = 8;
+CONSTs = cell(1,numfish);
+for i_fish = 1:numfish,
+    disp(['i_fish = ' num2str(i_fish)]);
+    
+    % for new partitioned data, consistant with GUI_050415, but not tested
+    fishdir = fullfile(data_dir,['CONST_F' num2str(new_i_fish) '_fast.mat']);
+    load(fishdir,'const');
+    load(fishdir,'dimCR');
+    CellResp = zeros(dimCR);
+    num = 0;
+    nParts = round(dimCR(1)*dimCR(2)/42000000);
+    for i = 1:nParts,
+        load(fishdir,['CRZt_' num2str(i)']);
+        eval(['len = size(CRZt_' num2str(i) ',1);']);
+        eval(['CellResp(num+1:num+len,:) = CRZt_' num2str(i) ';']);
+        eval(['CellResp(num+1:num+len,:) = CRZt_' num2str(i) ';']);
+        num = num+len;
+    end
+    names = fieldnames(const); % cell of strings
+    for i = 1:length(names),
+        % renaming exception!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
+%         if strcmp(names{i},'CRAZ'),
+%             setappdata(hfig,'CellRespAvr',const.CRAZ);
+%         elseif strcmp(names{i},'photostate'),
+%             setappdata(hfig,'stim_full',const.photostate);
+        else
+            setappdata(hfig,names{i},eval(['const.',names{i}]));
+        end
+    end
+    % add one variable
+    CONSTs{i_fish}.numcell = length(const.CInfo);
+    
+%%
+    % load selected clusters
+    CONSTs{i_fish}.clus = [];
+    CIX = [];
+    for i = 1, % just one for now...
+        if length(VAR(i_fish).ClusGroup{1})-i+1>0,
+            % copy from VAR
+            clus = VAR(i_fish).ClusGroup{1}(i);
+            CONSTs{i_fish}.clus(i).cIX = clus.cIX;
+            CONSTs{i_fish}.clus(i).gIX = clus.gIX;
+            CONSTs{i_fish}.clus(i).name = clus.name;
+            CONSTs{i_fish}.clus(i).numK = clus.numK;
+            CIX = [CIX; clus.cIX];
+        end
+    end
+    % copy from CONST
+    CIX = unique(CIX);
+    CONSTs{i_fish}.CIX = CIX;
+    CONSTs{i_fish}.CInfo = const.CInfo(CIX); % CIF
+    CONSTs{i_fish}.CRAZ = const.CRAZ(CIX,:);
+    CONSTs{i_fish}.CRZt = CellResp(CIX,:);
+    
+    %% OLD FORMAT without partitioning
+%     % load general info
+%     load(['CONST_F' num2str(i_fish) '.mat'],'CONST');
+%     % CONST: {'ave_stack','anat_yx','anat_yz','CInfo','periods','shift',...
+%     %     'CRAZ','CRZt','dshift_fc','FcAvr','Fc','photostate','tlists','datanames'};
+% %%
+%     % load fields from CONST, with names preserved
+%     names = {'anat_yx','anat_yz','anat_zx','periods','shift','dshift','FcAvr','Fc','photostate','tlists','datanames'};
+% 
+%     for i = 1:length(names),
+%         eval(['CONSTs{i_fish}.',names{i},' = CONST.',names{i},';']);
+%     end
+% 
+%     % add one variable
+%     CONSTs{i_fish}.numcell = length(CONST.CInfo);
+% %%
+%     % load selected clusters
+%     CONSTs{i_fish}.clus = [];
+%     CIX = [];
+%     for i = 1, % just one for now...
+%         if length(VAR(i_fish).ClusGroup{1})-i+1>0,
+%             % copy from VAR
+%             clus = VAR(i_fish).ClusGroup{1}(i);
+%             CONSTs{i_fish}.clus(i).cIX = clus.cIX;
+%             CONSTs{i_fish}.clus(i).gIX = clus.gIX;
+%             CONSTs{i_fish}.clus(i).name = clus.name;
+%             CONSTs{i_fish}.clus(i).numK = clus.numK;
+%             CIX = [CIX; clus.cIX];
+%         end
+%     end
+%     % copy from CONST
+%     CIX = unique(CIX);
+%     CONSTs{i_fish}.CIX = CIX;
+%     CONSTs{i_fish}.CInfo = CONST.CInfo(CIX); % CIF
+%     CONSTs{i_fish}.CRAZ = CONST.CRAZ(CIX,:);
+%     CONSTs{i_fish}.CRZt = CONST.CRZt(CIX,:);
+end
+%%
+clear CONST;
+temp = whos('CONSTs');
+if [temp.bytes]>2*10^9,
+    save('CONSTs_current.mat','CONSTs','-v7.3');
+else
+    save('CONSTs_current.mat','CONSTs');
+end
+
+%% optional
+timestamp  = datestr(now,'mmddyy_HHMM');
+currentfolder = pwd;
+save([currentfolder '\arc mat\CONSTs_' timestamp '.mat'],'CONSTs');
diff --git a/GUIpreload_step3_align.m b/GUIpreload_step3_align.m
new file mode 100644
index 0000000..864e55d
--- /dev/null
+++ b/GUIpreload_step3_align.m
@@ -0,0 +1,298 @@
+%%%%%%%%%%%%%%%
+%%
+% clear all;close all;
+% varList = {'CR_dtr','nCells','CInfo','anat_yx','anat_yz','anat_zx','ave_stack','fpsec','frame_turn','periods'};
+
+% % or
+varList = {'nCells','CInfo','anat_yx','anat_yz','anat_zx','ave_stack','fpsec','frame_turn'};
+
+%%
+M_dir = {'F:\Janelia2014\Fish1_16states_30frames';
+    'F:\Janelia2014\Fish2_20140714_2_4_16states_10frames';
+    'F:\Janelia2014\Fish3_20140715_1_1_16_states_10frames';
+    'F:\Janelia2014\Fish4_20140721_1_8_16states_20frames';
+    'F:\Janelia2014\Fish5_20140722_1_2_16states_30frames';
+    'F:\Janelia2014\Fish6_20140722_1_1_3states_30,40frames';
+    'F:\Janelia2014\Fish7_20140722_2_3_3states_30,50frames';
+    'F:\Janelia2014\Fish8_20141222_2_2_7d_PT_3OMR_shock_lowcut';
+    'F:\Janelia2014\Fish9_20150120_2_1_photo_OMR_prey_blob_blue_cy74_6d_20150120_220356';
+    'F:\Janelia2014\Fish10_20150120_2_2_photo_OMR_prey_blob_blue_cy74_6d_20150120_231917'};
+
+M_stimset = [1, 1, 1, 1, 1, 1, 1, 2, 2, 2];
+
+save_dir = 'F:\Janelia2014';
+
+%% MANUAL
+for i_fish = 10, %:8,    
+    disp(num2str(i_fish));
+    %% load data
+    datadir = M_dir{i_fish};
+    load(fullfile(datadir,['Fish' num2str(i_fish) '_direct_load.mat']),varList{:});
+    
+    %% ONLY RUN ONCE!!!!!!!!!
+    % add 1 frame at start:
+    CR_dtr = horzcat(CR_dtr(:,1),CR_dtr);
+
+    frame_turn = vertcat(frame_turn(1,:),frame_turn);
+    
+    % correction of an error in Fish #1
+    if i_fish == 1,
+        CR_dtr = horzcat(CR_dtr(:,1:20),CR_dtr);
+    end
+    
+    %% index processing
+    if M_stimset(i_fish)==1, % fish 1-7
+        M_period = {480,160,160,320,480,280,300}; %,{120,150,360}};
+        periods = M_period{i_fish};
+        period = periods;
+        
+        nrep = floor(size(CR_dtr,2)/period)-1;
+        dshift = 2;
+        shift = period+dshift;
+        
+        IX_all = 1+shift:period*nrep+shift;
+        CellResp = CR_dtr(:,IX_all); % old name: CRZt, 'Cell Responses Zscore trimmed'
+        
+        IX_avr = 1+shift:period+shift;
+        nCells = size(CR_dtr,1);
+        CRAZ = mean(reshape(CR_dtr(:,IX_all),nCells,period,[]),3); % 'Cell Response Average Zscore'
+        
+        datanames = {'rep average','all reps','rep #1','rep #2','last rep'};
+        ix_avr = 1:period;
+        ix_all = 1:period*nrep;
+        tlists = {ix_avr, ix_all};
+        for i = [1,2,nrep],
+            ix = 1+period*(i-1):period*i;
+            tlists = [tlists, ix];
+        end
+        
+        stim_full = frame_turn(17,IX_all);
+        stim_full = round(stim_full);
+        
+    else % multiple protocols
+        % process raw stimulus code
+        [stimset,stim_full] = StimulusKeyPreprocessing(frame_turn);
+        
+        % find time-lists = list of time-frames numbers for a certain stimulus
+        nTypes = length(stimset);
+        tlists_raw = cell(1,nTypes+1); % time-lists, i.e. frame indices
+        IX_all = [];
+        for i_type = 1:nTypes,
+            nSets = length(stimset(i_type).starts);
+            IX = [];
+            for i_set = 1:nSets,
+                IX = horzcat(IX,stimset(i_type).starts(i_set):stimset(i_type).stops(i_set));
+            end
+            tlists_raw{i_type} = IX;
+            IX_all = horzcat(IX_all,IX);
+        end
+        tlists_raw{nTypes+1} = IX_all;
+        
+        % This is the main data to store and load to GUI
+        CellResp = CR_dtr(:,IX_all); % old name: CRZt, 'Cell Responses Zscore trimmed'
+
+        %% get tlists corresponding to IX_all (tlists_raw ~ 1:totalframe#)
+        tlists = cell(1,nTypes+1); % time-lists, i.e. frame indices
+        for i_type = 1:nTypes,
+            [~,tlists{i_type}] = intersect(IX_all,tlists_raw{i_type});
+        end
+        
+        %% find average
+        CRAZ = []; % 'Cell Response Average Zscore'
+        stimtypelist = 1:nTypes;
+        numcell = size(CR_dtr,1);
+        for i = stimtypelist,
+            M = circshift(CR_dtr(:,tlists_raw{i}),shift,2);
+            CRAZ = horzcat(CRAZ,mean(reshape(M,numcell,periods(i),[]),3));
+        end
+        
+        %% variables to save later in struct
+        periods = [stimset.period];
+        datanames = {'rep average','all reps',stimset.name};                
+        
+        dshift = 2;
+        shift = -dshift; % circshift fluo left by 2 frames      
+        
+    end
+    
+    %% prepare fictive data
+    rows = [7,8,9,13,14];
+    F = frame_turn(:,rows)';
+    
+    F(2,:) = -F(2,:);
+    Fc = F(:,IX_all);
+        
+    % find averages
+    if M_stimset(i_fish)==1, % fish 1-7
+        FcAvr = mean(reshape(Fc,length(rows),period,[]),3);
+    else
+        FcAvr = []; % 'Cell Response Average Zscore'
+        stimtypelist = 1:nTypes;
+        numcell = size(CR_dtr,1);
+        for i = stimtypelist,
+            avr = mean(reshape(F(:,tlists_raw{i}),length(rows),periods(i),[]),3);
+            FcAvr = horzcat(FcAvr,avr);
+        end
+    end
+    
+    % normalizations
+    for i = 1:3,
+        m = FcAvr(i,:);
+        FcAvr(i,:) = (m-min(m))/(max(m)-min(m));
+        m = Fc(i,:);
+        Fc(i,:) = (m-min(m))/(max(m)-min(m));
+    end
+    m = FcAvr(4:5,:);
+    FcAvr(4:5,:) =  (m-min(min(m)))/(max(max(m))-min(min(m)));
+    m = Fc(4:5,:);
+    Fc(4:5,:) =  (m-min(min(m)))/(max(max(m))-min(min(m)));
+    
+    
+    %% compile CONST
+    CONST = [];
+    names = {'ave_stack','anat_yx','anat_yz','anat_zx','CInfo','periods','shift','CellResp','CRAZ','dshift','Fc','FcAvr','stim_full','stimset','tlists','datanames'};
+    for i = 1:length(names), % use loop to save variables into fields of CONST
+        eval(['CONST.',names{i},'=', names{i},';']);
+    end
+    
+    %% and save
+    %%% old method:
+    %     temp = whos('CONST');
+    %     if [temp.bytes]<2*10^9,
+    %         save(['CONST_F' num2str(i_fish) '.mat'],'CONST');
+    %         beep;
+    %     else
+    %         save(['CONST_F' num2str(i_fish) '.mat'],'CONST','-v7.3');
+    %         beep;
+    %     end
+    
+    %%% new method with partitioning of main data
+    newfishdir = fullfile(save_dir,['CONST_F' num2str(i_fish) '_fast.mat']);
+    const = CONST;
+    const = rmfield(const,'CellResp');
+    dimCR = size(CONST.CellResp);
+    save(newfishdir,'const','-v6');
+    save(newfishdir,'dimCR','-v6','-append');
+    
+    nParts = round(numel(CONST.CellResp)/42000000);
+    t = zeros(1,nParts);
+    ix = round(linspace(1,size(CONST.CellResp,1),nParts+1));
+    for i = 1:nParts,
+        disp(num2str(i));
+        eval(['CellResp_' num2str(i) '= CONST.CellResp(ix(i):ix(i+1),:);']);
+        save(newfishdir,['CellResp_' num2str(i)],'-v6','-append');
+    end
+
+
+    %% OR
+    
+end
+
+    %% initialize VAR (once)
+    
+
+%% Initialize VARS % outdated? Clusgroup?
+
+nCells = length(CONST.CInfo);
+
+    cIX = (1:nCells)';
+    i = 1;
+    VAR(i_fish).Class(i).name = 'all processed';
+    VAR(i_fish).Class(i).cIX = cIX;
+    VAR(i_fish).Class(i).gIX = ones(length(cIX),1);
+    VAR(i_fish).Class(i).numel = nCells;
+    VAR(i_fish).Class(i).numK = 1;
+    VAR(i_fish).Class(i).datatype = 'std';
+    
+    cIX = (1:100:nCells)';
+    VAR(i_fish).ClusGroup{1,1}.name = 'test';
+    VAR(i_fish).ClusGroup{1,1}.cIX = cIX;
+    VAR(i_fish).ClusGroup{1,1}.gIX = ones(length(cIX),1);
+    VAR(i_fish).ClusGroup{1,1}.numel = length(cIX);
+    VAR(i_fish).ClusGroup{1,1}.numK = 1;
+
+    %%
+        cIX = (1:10:nCells)';
+    VAR(i_fish).ClusGroup{1,1}(12).name = '1/10 of all';
+    VAR(i_fish).ClusGroup{1,1}(12).cIX = cIX;
+    VAR(i_fish).ClusGroup{1,1}(12).gIX = ones(length(cIX),1);
+    VAR(i_fish).ClusGroup{1,1}(12).numel = length(cIX);
+    VAR(i_fish).ClusGroup{1,1}(12).numK = 1;
+
+    %% varience/std for reps for each cell
+%     if i_fish==2 || i_fish==3 || i_fish==6,
+%         period_real = period/2;
+%     else
+%         period_real = period;
+%     end
+%     nrep_real = floor((size(CR,2)-shift)/period_real);
+%     while period_real*nrep_real+shift>size(CR,2),
+%         nrep_real = nrep_real-1;
+%     end
+%     CRZ_3D = reshape(CRZ(:,1+shift:period_real*nrep_real+shift),nCells,period_real,[]);
+%     %% updated method, weighing both std between each rep and (summed with) std of 1st half & 2nd half of experiment - 1/8/15
+%     % CRZ = CONST.M_array.CellResp;
+%     % if i_fish==2 || i_fish==3 || i_fish==6,
+%     %     period_real = CONST.M_array.period/2;
+%     % else
+%     %     period_real = CONST.M_array.period;
+%     % end
+%     % CRZ_3D = reshape(CRZ,size(CRZ,1),period_real,[]);
+%     % divide = round(size(CRZ_3D,3)/2);
+%     % CRZ_std1 = std(CRZ_3D(:,:,1:divide),0,3);
+%     % CRZ_std2 = std(CRZ_3D(:,:,divide+1:end),0,3);
+%     % temp1 = mean(CRZ_std1,2);
+%     % temp2 = mean(CRZ_std2,2);
+%     %
+%     % temp12 = horzcat(temp1,temp2);
+%     % temp = mean(temp12,2)+std(temp12,0,2);
+%     % [~,I] = sort(temp);
+%     % M = temp(I);
+%     % figure;plot(M)
+%     %
+%     % figure;imagesc(CRZ(I,:))
+%     %
+%     % nCells = size(CRZ,1);
+%     
+%     %% find low variance / stimulus-locked cells
+%     CRZ_std = std(CRZ_3D,0,3);
+%     temp = mean(CRZ_std,2);
+%     
+%     % find mean-std thres: 0.5
+%     [~,I] = sort(temp);
+%     M = temp(I);
+%     figure;plot(M)
+%     %%
+%     i_last = length(VAR(i_fish).Class);
+%     M_perc = [0.025,0.1,0.3];
+%     for j = 1:length(M_perc);
+%         thres = M(round(nCells*M_perc(j)));
+%         cIX = find(temp<thres);
+%         i = j+i_last;
+%         VAR(i_fish).Class(i).round = 0;
+%         VAR(i_fish).Class(i).name = ['perc < ' num2str(M_perc(j)*100) '%'];
+%         %     VAR(i_fish).Class(i).notes = ['perc < ' num2str(M_perc(j)*100) '%'];
+%         VAR(i_fish).Class(i).cIX = cIX;
+%         VAR(i_fish).Class(i).gIX = ones(length(cIX),1);
+%         VAR(i_fish).Class(i).numel = length(cIX);
+%         VAR(i_fish).Class(i).numK = 1;
+%         VAR(i_fish).Class(i).datatype = 'std';
+%     end
+%     
+%     %% shift CR?
+%     % shift = 161;
+%     % nrep=floor(size(CR,2)/period)-1;
+%     %
+%     % skiplist=[];
+%     % IX_rep=setdiff(1:nrep, skiplist);
+%     % IX=zeros(period*length(IX_rep),1);
+%     %
+%     % for i=1:length(IX_rep)
+%     %     IX(period*(i-1)+1:period*i)=period*(IX_rep(i)-1)+1+shift:period*(IX_rep(i))+shift;
+%     % end
+%     %
+%     % % cell_resp=cell_resp(:,1:nrep*period);
+%     % CRA=mean(reshape(CR(:,IX),[nCells,period,length(IX_rep)]),3);
+%     % CRAZ = zscore(CRA')';
+%     
+
diff --git a/GUIpreload_step4_loadGUI.m b/GUIpreload_step4_loadGUI.m
new file mode 100644
index 0000000..6ee11ac
--- /dev/null
+++ b/GUIpreload_step4_loadGUI.m
@@ -0,0 +1,30 @@
+% Instant Load!
+
+% change current folder to the folder containing this m-file
+
+clear all;close all;clc
+
+global VAR;
+load('VAR_current.mat','VAR');
+
+load('CONSTs_current.mat','CONSTs');
+
+data_dir = [pwd '\example data']; % 'F:\Janelia2014';%
+
+%% Start GUI
+
+[hfig,fcns] = GUI_FishExplorer(data_dir,CONSTs,VAR);
+
+% to push new functions to GUI, called with function 'push_cIX_gIX':
+% hfig = GUI_FishExplorer(data_dir,CONSTs,VAR,flag_script,var_script)
+
+% to get handle of local functions from GUI to use in workspace 
+% (call by using 'f.FunctionName')
+f = [];
+for i = 1:length(fcns),
+    eval(['f.' char(fcns{i}) ' = fcns{i};']);
+end
+
+%%
+% to recover after closing figure:
+% global EXPORT_autorecover;
diff --git a/PartitionForFastSaveLoad.m b/PartitionForFastSaveLoad.m
new file mode 100644
index 0000000..651fc76
--- /dev/null
+++ b/PartitionForFastSaveLoad.m
@@ -0,0 +1,85 @@
+
+%% Partition test
+nParts = 8;
+t = zeros(1,nParts);
+ix = round(linspace(1,size(CONST.CRZt,1),nParts+1));
+for i = 1:nParts,
+    tic;
+    eval(['CRZt_' num2str(i) '= CONST.CRZt(ix(i):ix(i+1),:);']);
+    save('CONST_F9_multi.mat',['CRZt_' num2str(i)'],'-v6','-append');
+    t(i) = toc;
+    i
+    t(i)
+end
+sum(t)
+
+%% reconstruct
+tic
+
+CRZt = zeros(size(CONST.CRZt));
+num = 0;
+for i = 1:8,
+    eval(['len = size(CRZt_' num2str(i) ',1);']);
+    eval(['CRZt(num+1:num+len,:) = CRZt_' num2str(i) ';']);
+    num = num+len;
+end
+
+toc
+
+%% for reals: partition and save
+tic
+const = CONST;
+const = rmfield(const,'CRZt');
+save('CONST_F9_fast.mat','const','-v6');
+
+nParts = 8;
+t = zeros(1,nParts);
+ix = round(linspace(1,size(CONST.CRZt,1),nParts+1));
+for i = 1:nParts,
+    disp(num2str(i));
+    eval(['CRZt_' num2str(i) '= CONST.CRZt(ix(i):ix(i+1),:);']);
+    save('CONST_F9_fast.mat',['CRZt_' num2str(i)'],'-v6','-append');
+end
+toc
+
+%% load
+tic
+load('CONST_F9_fast.mat','const');
+load('CONST_F9_fast.mat','dimCR');
+CRZt = zeros(dimCR);
+num = 0;
+for i = 1:8,
+    load('CONST_F9_multi.mat',['CRZt_' num2str(i)']);
+    eval(['len = size(CRZt_' num2str(i) ',1);']);
+    eval(['CRZt(num+1:num+len,:) = CRZt_' num2str(i) ';']);
+    num = num+len;
+end
+toc
+
+%% reformat old mats
+tic
+data_dir = 'F:\Janelia2014';
+for i_fish = 2:8,
+    disp(['fish ' num2str(i_fish)]);
+    fishdir = fullfile(data_dir,['CONST_F' num2str(i_fish) '.mat']);
+    newfishdir = fullfile(data_dir,['CONST_F' num2str(i_fish) '_fast.mat']);
+    load(fishdir,'CONST');
+    const = CONST;
+    const = rmfield(const,'CRZt');
+    dimCR = size(CONST.CRZt);
+    save(newfishdir,'const','-v6');
+    save(newfishdir,'dimCR','-v6','-append');
+    
+    nParts = round(numel(CONST.CRZt)/42000000);
+    t = zeros(1,nParts);
+    ix = round(linspace(1,size(CONST.CRZt,1),nParts+1));
+    for i = 1:nParts,
+        disp(num2str(i));
+        eval(['CRZt_' num2str(i) '= CONST.CRZt(ix(i):ix(i+1),:);']);
+        save(newfishdir,['CRZt_' num2str(i)],'-v6','-append');
+    end
+
+end
+toc
+
+
diff --git a/README.txt b/README.txt
new file mode 100644
index 0000000..f811499
--- /dev/null
+++ b/README.txt
@@ -0,0 +1,20 @@
+To start:
+Open "GUIpreload_step4_loadGUI.m" 
+Change Matlab's current folder to folder containing this m-file.
+Run this file and GUI should start.
+
+Main code is in "GUI_FishExplorer.m".
+
+For most fish, only data from 2 clusters are saved into "CONSTs_current.mat".
+Only full data for one example fish, Fish 6, is included. 
+To view data of full fish, under tab "General", select "6" in drop-down menu "Load full fish". Then most functions can be used. 
+
+To see some example selections of cells, go to tab "Saved Clusters", and browse drop-down menus "Class" and "Clusters".
+
+In tab "Operations", use the "Back" and "Forward" button to access the previous views. "Choose cluster range" takes inputs like "1-3,5" or "2:end". Cluster range are indicated by the numbers next to the color bar. 
+
+For descriptions of the other functions, please read the description at the top of "GUI_FishExplorer.m".
+
+Please let me know if you have any questions!
+
+- Xiuye Chen (xiuyechen@gmail.com)
diff --git a/StimulusKeyPreprocessing.m b/StimulusKeyPreprocessing.m
new file mode 100644
index 0000000..9ce02d7
--- /dev/null
+++ b/StimulusKeyPreprocessing.m
@@ -0,0 +1,401 @@
+function [stimset,stim] = StimulusKeyPreprocessing(frame_turn)
+
+isPlotting = 1; % plot parsing process for manual inspection!
+
+% Manual inputs needed for this function:
+% - protocol sequence and naming, as in 'block_raw' and 'stimset' initialization
+% - substitution array 'M_range', to substitute original stimulus code to a
+% tighter range that is better defined for plotting (both for plotting values and stimulus bar)
+
+%% Raw Stimlus Key ----------------------------------------------------------------------------- MANUAL INPUT!
+
+% Manually input protocol sequence
+% stored in 'block', based on the raw stimulus code
+%% Manual: params for Fish 8
+nBlocks = 3;
+block_raw = cell(nBlocks,1); % number of blocks
+% write out protocol sequence for each block; write stimulus sets within block in arrays within cell
+block_raw{1} = {[2,3,4,5],[12,13,12,14,12,15],99,};
+block_raw{2} = {[2,100,2,3,4,100,4,5],[2,100,2,3,4,100,4,5],[12,100,12,13,12,100,12,14,12,100,12,15],99};
+block_raw{3} = {[2,3,4,5],[12,13,12,14,12,15],99};
+
+stimset = [];
+stimset(1).name = 'PT (blue/red)';
+stimset(1).ij = [1,1; 3,1;];
+stimset(2).name = 'OMR';
+stimset(2).ij = [1,2; 3,2;];
+stimset(3).name = 'Spontaneous';
+stimset(3).ij = [1,3; 2,4;3,3];
+
+%%
+% PT/Color(Blue/Red)
+% 2,4: Red|Red
+% 3: Blue|Red
+% 5: Red|Blue
+%
+% OMR
+% 12: Red
+% 13: OMR F
+% 14: OMR R
+% 15: OMR L
+%
+% Spont
+% 99: same as black
+%
+% Dot(prey)
+% 23: No dot
+% 22: Dot
+%
+% Blob
+% 30: Red
+% 31: Blob R
+% 32: Blob L
+
+
+%% Manual: params for Fish 10
+% block_raw{1} = {[2,3,4,5],[12,13,12,14,12,15],99,[23,22],[30,31,30,32],[2,3,4,5]};
+% block_raw{2} = {[2,100,2,3,4,100,4,5],[12,100,12,13,12,100,12,14,12,100,12,15],99,[23,22],[30,31,30,32],[2,100,2,3,4,100,4,5]};
+% block_raw{3} = {[2,3,4,5],[12,13,12,14,12,15],99,[23,22],[30,31,30,32],[2,3,4,5],[12,13,12,14,12,15]};
+% 
+% stimset = [];
+% stimset(1).name = 'PT (blue/red)';
+% stimset(1).ij = [1,1; 1,6; 3,1; 3,6];
+% stimset(2).name = 'OMR';
+% stimset(2).ij = [1,2; 3,2; 3,7];
+% stimset(3).name = 'Spontaneous';
+% stimset(3).ij = [1,3; 2,3; 3,3];
+% stimset(4).name = 'Dot/prey';
+% stimset(4).ij = [1,4; 2,4; 3,4];
+% stimset(5).name = 'Blob';
+% stimset(5).ij = [1,5; 2,5; 3,5];
+
+%% Load stimulus
+st = frame_turn(:,17)'; % from raw file
+
+% cap abnormally high values in raw values
+temp = sort(st,'descend');
+thr = temp(round(length(st)/50)); % arb. thr
+st(st>thr) = thr+1; % assign cap value to be thr+1
+
+xv = 1:thr+1; % x-vector for histogram bins
+
+% Manual inspection
+% figure;
+% hist(st,xv);
+
+% (continue by default) find bins with significantly high counts, i.e.
+% whole range/set of raw stimulus keys, store in 'M_range_raw'
+% use to manually define standardized 'M_range'
+counts = hist(st,xv);
+thr_counts = round(length(st)/500);
+M_range_raw = xv(counts>thr_counts);
+disp(['M_range_raw = ' num2str(M_range_raw)]); % use this to adjust the substitution array, M_range, manually after this inspection!%%%%%%%%%%%%%%%%%%%%%%
+
+%% Standardized Stimulus Key ----------------------------------------------------------------------------- MANUAL INPUT!
+% Define normalized code, e.g. same stimulus under different raw stim keys
+% can be unified
+M_range =    [3,1,3,2, 9,10,11,12, 0,13, 3,14,15, 4,16]; % corresponding to
+% raw      = [2,3,4,5,12,13,14,15,22,23,30,31,32,99,100] % for Fish 10
+
+% 0 = all black; 1 = black/white; 2 = white/black; 3 = all white; 4 = all gray;
+% 5 = gray/black; 6 = white/gray; 7 = black/gray; 8 = gray/white.
+% OMR baseline = not moving?? grey??
+% 10 = forward grating (very slow, more for calibration)
+% 11 = rightward grating
+% 12 = leftward grating
+
+
+% half-fields can be displayed directly. other L/R directional cues can be
+% displayed by gradient. If not-directional or unspecified, will assign a
+% color-code randomly.
+
+%% correct transient frames
+% short method: st_rounded = interp1(M_range,M_range,st,'nearest');
+% but this doesn't work if a frame is different from both the one
+% before and the one after. Use loop below to force a manual 'round'
+% to the closer one of the 2 neighbors.
+temp = diff([0,st]);
+switches = find(temp);
+trans = switches(diff(switches)==1);
+
+for i = 1:length(trans),
+    j = trans(i);
+    if j>1,
+        st(j) = st(j-1) + round((st(j)-st(j-1))/abs(st(j+1)-st(j-1))); % round to nearest of st(j-1) and st(j+1)
+    end
+end
+
+% (loop until all transient ones are corrected)
+temp = diff([0,st]);
+switches = find(temp);
+trans = switches(diff(switches)==1);
+count = 0;
+while ~isempty(trans),
+    for i = 1:length(trans),
+        j = trans(i);
+        st(j) = st(j-1) + round((st(j)-st(j-1))/abs(st(j+1)-st(j-1))); % round to nearest of st(j-1) and st(j+1)
+    end
+    temp = diff([0,st]);
+    switches = find(temp);
+    trans = switches(diff(switches)==1);
+    count = count+1;
+    if count>5,
+        disp('count>5??');
+        break;
+    end
+end
+
+%% Standardize stimulus keys for both 'stim' and 'block', based on 'M_range_raw'->'M_range'
+% (tricky for cell arrays) custom method:
+% make a replacement matrix 'M_replace' to substitude raw stimulus key in 'block_raw'
+%  to normalized key, stored in new cell array 'block'
+M_replace = zeros(1,max(M_range_raw));
+for i = 1:max(M_range_raw),
+    ix = find(i==M_range_raw);
+    if ~isempty(ix),
+        M_replace(i) = M_range(ix);
+    end
+end
+
+stim = zeros(size(st));
+block = cell(nBlocks,1);
+for i = 1:length(M_range_raw),
+    stim(st==M_range_raw(i)) = M_range(i);
+    for j = 1:nBlocks,
+        block{j} = cellfun(@(x) M_replace(x),block_raw{j},'UniformOutput',0);
+    end
+end
+
+%% plot stimulus-key!
+
+
+%% Stimulus set segmentation, based on 'block' (manual input of protocol sequence)
+% reduce 'stim' to a sequence of keys without consecutive repeats
+ix_singles = find(diff([0,stim])); % index array to map back to raw indices (frame-number)
+singles = stim(ix_singles); % sequence of keys without consecutive repeats.
+% '_sg' below stands for 'singles'. Manipulations below in both raw and singles, in parallel.
+
+jump = 1; % searching for 'jumps' in 'singles'
+% Based on the known protocol sequence for each set, look for first
+% mismatch in 'singles', i.e. position where the next set begins, stored
+% as 'block_change(_sg)' and 'set_start(_sg)'.
+block_change = cell(nBlocks,1);
+block_change_sg = cell(nBlocks,1);
+for I = 1:nBlocks,
+    nSets = length(block{I});
+    block_change{I} = zeros(1,nSets);
+    block_change_sg{I} = zeros(1,nSets);
+    for J = 1:nSets,
+        jump_last = jump;
+        pattern = block{I}{J}; % protocol sequence for this set
+        
+        % this is sort of awkward. Instead of padding the 'already-found' sets with zeros,
+        % one could also search the cropped array and adjusted the index.
+        % Now 'comparison' is aligned with stim, works.
+        textile = repmat(pattern,1,ceil(length(singles)/length(pattern))); % tile, for use below
+        
+        template = zeros(size(singles));
+        ixs_fill = jump_last:length(singles);
+        template(ixs_fill) = textile(1:length(ixs_fill)); % crop the 'textile' to fill the space after jump_last
+        
+        singles_temp = singles;
+        if jump_last>1,
+            singles_temp(1:jump_last-1) = 0;
+        end
+        comparison = (singles_temp-template);
+        % find the position where the next set begins
+        jump = find((comparison),1,'first');
+        
+        %% contingency plan: if start of the new set is different from
+        % expected protocol ('block')
+        count_stuck = 0;
+        count_stuck2 = 0;
+        while jump-jump_last<length(pattern),
+            count_stuck = count_stuck+1;
+            if count_stuck>length(pattern),
+                count_stuck = 0;
+                jump_last = jump_last+1;
+                count_stuck2 = count_stuck2+1;
+                disp(['count_stuck2=' num2str(count_stuck2)]);
+                if count_stuck2>10, % arb. thres
+                    disp('count_stuck2>10');
+                    break;
+                end
+            end
+            
+            pattern_ = circshift(pattern,-count_stuck,2);
+            textile = repmat(pattern_,1,ceil(length(singles)/length(pattern_)));
+            
+            template = zeros(size(singles));
+            ixs_fill = jump_last:length(singles);
+            template(ixs_fill) = textile(1:length(ixs_fill));
+            
+            singles_temp = singles;
+            if jump_last>1,
+                singles_temp(1:jump_last-1) = 0;
+            end
+            comparison = (singles_temp-template);
+            
+            jump = find((comparison),1,'first');
+            
+        end
+        %% save
+        if isempty(jump), % reached end of experiment
+            %%
+            % NOTE END OF
+            % EXPERIMENT????????????????????????????????????????????????????????????????????????
+            break;
+        end
+        block_change_sg{I}(J) = jump;
+        block_change{I}(J) = ix_singles(jump);
+    end
+end
+
+% eliminate empty sets
+for I = 1:nBlocks,
+    nSets = length(block_change{I});
+    for J = 1:nSets,
+        if block_change{I}(J) == 0,
+            block_change{I}(J:end) = [];
+            block_change_sg{I}(J:end) = [];
+            if length(block{I})>= J,
+                block{I}(J:end) = [];
+            end
+            break;
+        end
+    end
+end
+
+% block_change is basically start of next set. shift by one set to get 'set_start'.
+set_start = cell(nBlocks,1);
+set_start_sg = cell(nBlocks,1);
+set_stop = cell(nBlocks,1);
+set_stop_sg = cell(nBlocks,1);
+for I = 1:nBlocks,
+    nSets = length(block{I});
+    set_start{I} = zeros(1,nSets);
+    set_start_sg{I} = zeros(1,nSets);
+    if I == 1,
+        set_start{I}(1) = 1;
+        set_start_sg{I}(1) = 1;
+    else
+        set_start{I}(1) = block_change{I-1}(end);
+        set_start_sg{I}(1) = block_change_sg{I-1}(end);
+    end
+    set_start{I}(2:end) = block_change{I}(1:end-1);
+    set_start_sg{I}(2:end) = block_change_sg{I}(1:end-1);
+    set_stop{I} = block_change{I}-1;
+    set_stop_sg{I} = block_change_sg{I}-1;
+end
+
+%% Visualize current segmentation of blocks
+if isPlotting,
+    figure;
+    hold on;
+    plot(stim)
+    for I = 1:nBlocks,
+        for J = 1:length(block_change{I}),
+            x = block_change{I}(J);
+            plot([x,x],[0,20],'r--')
+        end
+    end
+end
+%% Find shift-corrections for each stim-set, used for averaging later
+
+
+%% Find periods for each set-stype, and organize info by type into 'stimset'.
+for i_ss = 1:length(stimset),
+    % find empty sets (for this particular experiment) and delete ij's in 'stimset'
+    for i_SetRep = 1:size(stimset(i_ss).ij,1),
+        I = stimset(i_ss).ij(i_SetRep,1);
+        J = stimset(i_ss).ij(i_SetRep,2);
+        if I>length(block) || J>length(block{I}),
+            stimset(i_ss).ij(i_SetRep:end,:) = [];
+            break;
+        end
+    end
+    nSetReps = length(stimset(i_ss).ij);
+    
+    %% get period from first set of block
+    I = stimset(i_ss).ij(1,1);
+    J = stimset(i_ss).ij(1,2);
+    pattern = block{I}{J}; % protocol sequence (singles) in normalized code
+    stimset(i_ss).pattern = pattern;
+    
+    % find periodicity
+    if length(pattern)==1, % exception, e.g. 'Spontaneous' set, single stimlus key held for full duration of set
+        if J < length(block{I}),
+            next_start = set_start{I}(J+1);
+        elseif J == length(block{I}),
+            next_start = set_start{I+1}(1);
+        else
+            next_start = length(stim);
+        end
+        stimset(i_ss).period = next_start - set_start{I}(J);
+    else % find periodicity based on stimlus key repetition
+        for i = 1:length(pattern),
+            if length(find(pattern==pattern(i)))==1, % use a stimulus that appears only once per period
+                % find period
+                start_s = set_start_sg{I}(J) + i -1;
+                stop_s = start_s + find(singles(1,start_s+1:end)==pattern(i),1,'first');
+                stimset(i_ss).period = ix_singles(stop_s)-ix_singles(start_s);
+                break;
+            end
+        end
+    end
+    
+    %%
+    %     stimset(i_ss).period % assigned above
+    
+    stimset(i_ss).rawstarts = zeros(1,nSetReps);
+    stimset(i_ss).rawstops = zeros(1,nSetReps);
+    stimset(i_ss).starts = zeros(1,nSetReps);
+    stimset(i_ss).stops = zeros(1,nSetReps);
+    stimset(i_ss).nReps = zeros(1,nSetReps);
+    for i_SetRep = 1:nSetReps,
+        I = stimset(i_ss).ij(i_SetRep,1);
+        J = stimset(i_ss).ij(i_SetRep,2);
+        stimset(i_ss).rawstarts(i_SetRep) = set_start{I}(J);
+        stimset(i_ss).rawstops(i_SetRep) = block_change{I}(J)-1;
+        
+        if length(pattern)==1, % exception, e.g. 'Spontaneous' set, single stimlus key held for full duration of set
+            stimset(i_ss).starts(i_SetRep) = stimset(i_ss).rawstarts(i_SetRep);
+            stimset(i_ss).stops(i_SetRep) = stimset(i_ss).rawstops(i_SetRep);
+            stimset(i_ss).nReps(i_SetRep) = 1;
+        else            
+            % circshift to find intact periods
+            pattern = stimset(i_ss).pattern;
+            
+            ix_start = set_start_sg{I}(J);
+            ix_stop = set_stop_sg{I}(J);
+            
+            % find 'shift' (shift = 1 means starting from 2nd in 'singles')
+            segment = singles(ix_start:ix_start+length(pattern)-1);
+            shift = 0;
+            while ~isequal(segment,pattern),
+                shift = shift+1;
+                segment = singles(ix_start+shift:ix_start+length(pattern)-1+shift);
+            end
+            stimset(i_ss).starts(i_SetRep) = ix_singles(ix_start+shift);
+            % round to period within this set
+            nReps = floor((ix_stop-ix_start-shift+1)/length(pattern));
+            stimset(i_ss).nReps(i_SetRep) = nReps;
+            
+            ix = stimset(i_ss).starts(i_SetRep) + nReps*stimset(i_ss).period - 1;
+            if ix>stimset(i_ss).rawstops(i_SetRep),
+                nReps = nReps-1;
+            end
+            stimset(i_ss).stops(i_SetRep) = stimset(i_ss).starts(i_SetRep) + nReps*stimset(i_ss).period - 1;
+        end
+    end
+end
+
+% discard very first period because of potential artifacts at beginning of
+% entire experiment
+if stimset(1).nReps(1)>1,
+    stimset(1).nReps(1) = stimset(1).nReps(1) - 1;
+    stimset(1).starts(1) = stimset(1).starts(1) + stimset(1).period;
+end
+
+
+end
\ No newline at end of file
diff --git a/arrows.jpg b/arrows.jpg
new file mode 100644
index 0000000..6d1147b
Binary files /dev/null and b/arrows.jpg differ
diff --git a/arrows_wide.jpg b/arrows_wide.jpg
new file mode 100644
index 0000000..9438967
Binary files /dev/null and b/arrows_wide.jpg differ
diff --git a/first version/AccessLocalFunctions.m b/first version/AccessLocalFunctions.m
new file mode 100644
index 0000000..1953044
--- /dev/null
+++ b/first version/AccessLocalFunctions.m	
@@ -0,0 +1,12 @@
+
+
+f = [];
+for i = 1:length(fcns),
+    eval(['f.' char(fcns{i}) ' = fcns{i};']);
+end
+
+%% test function 'FindCentroid'
+M = getappdata(hfig,'M');
+gIX = getappdata(hfig,'gIX');
+C_test = f.FindCentroid(gIX,M);
+figure;imagesc(C_test)
\ No newline at end of file
diff --git a/first version/BasicPlotMaps.m b/first version/BasicPlotMaps.m
new file mode 100644
index 0000000..9196801
--- /dev/null
+++ b/first version/BasicPlotMaps.m	
@@ -0,0 +1,261 @@
+function BasicPlotMaps(cIX,gIX,M,CInfo,photostate,anat_yx,anat_yz,anat_zx) %(cIX,gIX,cIX_0,M_0,CIF,numK,M)
+% M = M_0(cIX_0(cIX),:);
+numK = length(unique(gIX));
+%%
+dataFR = 1;
+figure('Position',[100 50 1350 900]);%,'DeleteFcn',@closefigure_Callback);
+h1 = axes('Position',[0.05, 0.06, 0.5, 0.85]); % left ~subplot
+BasicDrawClusters(h1,M,gIX,dataFR,numK,photostate);
+
+h2 = axes('Position',[0.58, 0.06, 0.40, 0.85]); % right ~subplot
+DrawClustersOnMap_LSh(CInfo,cIX,gIX,numK,anat_yx,anat_yz,anat_zx,'hsv');
+% set(gcf,'PaperUnits', 'inches', 'PaperSize', [12, 6])
+
+end
+
+function BasicDrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite)
+pos = get(gca,'Position');
+barratio = 0.02;
+numK = double(max(double(numK),double(max(gIX))));
+
+% sort traces by index
+im = M;
+[nLines,nFrames] = size(im);
+[~,I] = sort(gIX);
+im = im(I,:);
+
+%% convert imagesc effect into rgb matrix 'RGB'
+if dataFR, % is drawing cell-response fluorescence
+    cmap = gray(64);
+    im = AutoScaleImage0to1(im); % scaling to min/max 0/1
+    minlim = 0;
+    maxlim = 1;
+else % create blue-white-red map, for regression results
+    cmap = zeros(64,3);
+    cmap(:,1) = [linspace(0,1,32), linspace(1,1,32)];
+    cmap(:,2) = [linspace(0,1,32), linspace(1,0,32)];
+    cmap(:,3) = [linspace(1,1,32), linspace(1,0,32)];
+    minlim = -1; %min(min(im));
+    maxlim = 1; %max(max(im));
+end
+RGB = ImageToRGB(im,cmap,minlim,maxlim); % map image matrix to range of colormap
+
+%% add vertical color code
+if exist('clrmap','var'),
+    if strcmp(clrmap,'jet'),
+        temp = flipud(jet(numK));
+    else % 'hsv'
+        temp = hsv(round(numK*1.1));
+    end
+else % 'hsv'
+    temp = hsv(round(numK*1.1));
+end
+cmap2 = vertcat(temp(1:numK,:),[0,0,0]); % extend colormap to include black
+bwidth = max(round(nFrames/30),1);
+
+idx = gIX(I);
+ix_div = [find(diff(idx));length(idx)];
+
+bars = ones(nLines,bwidth);
+if numK>1,
+    bars(1:ix_div(1),:) = idx(ix_div(1));
+    for i = 2:length(ix_div),
+        % paint color
+        bars(ix_div(i-1)+1:ix_div(i),:) = idx(ix_div(i));
+    end
+end
+im_bars = reshape(cmap2(bars,:),[size(bars),3]);
+% add a white division margin
+div = ones(nLines,round(bwidth/2),3);
+% put 3 parts together
+im = horzcat(RGB,div,im_bars);
+
+%% plot figure
+% h_fig = figure('Position',[100 0 1300 900],'Name',['round ' num2str(i_step)]); %[left, bottom, width, height]
+[s1,s2,s3] = size(im);
+
+hold off;
+set(h1,'Position',[pos(1),pos(2)+pos(4)*barratio*3,pos(3),pos(4)*(1-4*barratio)]); 
+if s1<30,
+    temp = im;
+    im = ones(30,s2,s3);
+    im(1:s1,:,:) = temp;
+end
+image(im);
+set(gca, 'box', 'off')
+
+hold on;
+
+% plot cluster division lines
+if numK>1,
+    for i = 1:length(ix_div),% = numK-1,
+        y = ix_div(i)+0.5;
+        plot([0.5,s2+0.5],[y,y],'k','Linewidth',0.5);
+%         plot([1,s2*kx],[y,y],'k','Linewidth',0.5);
+    end
+end
+
+ylabel(['ROI number: ' num2str(s1)]);
+set(gca,'YTick',[],'XTick',[]);
+set(gcf,'color',[1 1 1]);
+
+colormap gray;
+
+% write in number label
+x = s2*1.003;
+y_last = 0;
+for i = 1:length(ix_div),
+    % avoid label crowding
+    margin = 0.015*s1;
+    y0 = ix_div(i)+0.5;
+    y = max(y_last+margin,y0);
+    if i<length(ix_div),
+        ynext = ix_div(i+1);
+    else
+        ynext = y0+margin*2;
+    end
+    % draw if not squeezed too much away
+    if y < y0+margin*2 && y < 0.5*(y0+ynext), %nLines+stimheight-margin,
+        if exist('iswrite','var') && iswrite,
+            try
+                text(x,y,[num2str(idx(ix_div(i))) ': ' num2str(rankscore(i))],'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
+                    'FontUnits','normalized','FontSize',0.015,'Tag','label');
+            catch
+                disp('rankscore invalid');
+                text(x,y,num2str(idx(ix_div(i))),'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
+                    'FontUnits','normalized','FontSize',0.015,'Tag','label');
+            end
+            
+        else
+            text(x,y,num2str(idx(ix_div(i))),'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
+                'FontUnits','normalized','FontSize',0.015,'Tag','label');
+        end
+        y_last = y;
+    end
+end
+
+if exist('stim','var'),
+    barlength = size(im,2);
+    DrawBars(pos,barratio,nFrames,barlength,stim);
+    
+    if exist('fictive','var'),
+        % Draw fictive
+        axes('Position',[pos(1),pos(2),pos(3),pos(4)*barratio*2]);
+        temp = ones(size(fictive,1),barlength);
+        temp(:,1:length(fictive)) = fictive;
+        imagesc(temp);colormap hot
+        axis off;
+    end
+end
+
+end
+
+%% Internal Functions
+
+function im = AutoScaleImage0to1(im)
+temp = mat2gray(im);
+low_high = stretchlim(temp,[0.005,0.995]);
+temp(temp<low_high(1)) = low_high(1);
+temp(temp>low_high(2)) = low_high(2);
+im = mat2gray(temp);
+end
+
+function RGB = ImageToRGB(im,cmap,minlim,maxlim)
+L = size(cmap,1);
+ix = round(interp1(linspace(minlim,maxlim,L),1:L,im,'linear','extrap'));
+RGB = reshape(cmap(ix,:),[size(ix) 3]); % Make RGB image from scaled.
+end
+
+function DrawBars(pos,barratio,nFrames,barlength,photostate) % horizontal stimulus bar
+% pos = get(gca,'Position'); % [0.05, 0.06, 0.5, 0.85]
+
+% global htop hbottom;
+barheight = 100;
+stimheight = 2*barheight;
+
+m1 = 0.3*ones(barheight,length(photostate));
+m2 = m1; % bottom half
+x = photostate;
+    
+%     2,3,4,5,12,13,14,15,99
+%     3,1,3,2,4 ,10,11,12,
+    
+if max(x)<=3,
+    % 0 = all black; 1 = black/white; 2 = white/black; 3 = all white; (4 = all gray;)
+    % 5 = gray/black; 6 = white/gray; 7 = black/gray; 8 = gray/white.
+    
+    % top (~ projection right)
+    m1(:,x==0 | x==2 | x==5) = 0; % black
+    m1(:,x==4 | x==6 | x==7) = 0.5; % grey
+    m1(:,x==1 | x==3 | x==8) = 1; % white
+    % bottom (~ projection left)
+    m2(:,x==0 | x==1 | x==7) = 0;
+    m2(:,x==4 | x==5 | x==8) = 0.5;
+    m2(:,x==2 | x==3 | x==6) = 1;
+    
+    temp = repmat(vertcat(m1,m2),[1 1 3]); % 3 color layers in dim3
+else
+    %%
+    % 2 = white/white
+    % 3 = black/white
+    % 4 = white/white
+    % 5 = white/black
+    %
+    % 12 = gray
+    % 13 = forward grating (very slow, more for calibration)
+    % 14 = rightward grating
+    % 15 = leftward grating  (* I hope I got right/left correct - if you think it's flipped it probably is)
+    %
+    % 99 = spontaneous
+    % top (~ projection right)
+    
+    m1(:,x==3) = 0; % black
+    m1(:,x==2 | x==4 | x==5) = 1; % white
+    % bottom (~ projection left)
+    m2(:,x==5) = 0;
+    m2(:,x==2 | x==3 | x==4) = 1;
+    
+    temp = repmat(vertcat(m1,m2),[1 1 3]); % 3 color layers in dim3
+    %%
+    temp(:,x==14,1) = 1; % red
+    temp(:,x==13,2) = 1; % green
+    temp(:,x==15,3) = 1; % blue
+    
+    % grey
+    temp(:,x==12,1) = 0.8;
+    temp(:,x==12,2) = 0.8;
+    temp(:,x==12,3) = 0.8;
+    
+    % yellow
+    temp(:,x==99,1) = 1;
+    temp(:,x==99,2) = 1;
+    temp(:,x==99,3) = 0.6;
+end
+
+n = ceil(nFrames/length(photostate)); % tile, to size or bigger
+temp = repmat(temp,1,n);
+stimbar = ones(stimheight,barlength,3);
+stimbar(:,1:nFrames,:) = temp(:,1:nFrames,:); % crop to size
+
+% if isempty(htop),
+axes('Position',[pos(1),pos(2)+pos(4)*(1-barratio),pos(3),pos(4)*barratio]);
+% elseif ~ishandle(htop),
+%     htop = axes('Position',[pos(1),pos(2)+pos(4)*(1-barratio),pos(3),pos(4)*barratio]);
+% else axes(htop)
+% end
+image(stimbar);axis off;hold on;
+% plot top border
+plot([1,length(photostate)],[1,1],'k','Linewidth',0.5);
+
+% if isempty(hbottom),
+axes('Position',[pos(1),pos(2)+pos(4)*barratio*2,pos(3),pos(4)*barratio]);
+% elseif ~ishandle(hbottom),
+%     hbottom = axes('Position',[pos(1),pos(2),pos(3),pos(4)*barratio]);
+% else axes(hbottom)
+% end
+
+image(stimbar);axis off;hold on;
+% plot bottom border
+plot([1,length(photostate)],[stimheight,stimheight],'k','Linewidth',0.5);
+%%
+end
diff --git a/Conv_test.m b/first version/Conv_test.m
similarity index 100%
rename from Conv_test.m
rename to first version/Conv_test.m
diff --git a/DA_2_Solution.m b/first version/DA_2_Solution.m
similarity index 100%
rename from DA_2_Solution.m
rename to first version/DA_2_Solution.m
diff --git a/first version/DrawClusters.m b/first version/DrawClusters.m
new file mode 100644
index 0000000..964da88
--- /dev/null
+++ b/first version/DrawClusters.m	
@@ -0,0 +1,210 @@
+function DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite)
+pos = get(gca,'Position');
+barratio = 0.03;
+
+%% Prepare cluster data
+% down-sample
+displaymax = 1000;
+numcell = size(M,1);
+if numcell > displaymax,
+    skip = round(numcell/displaymax);
+    M = M(1:skip:end,:);
+    gIX = gIX(1:skip:end,:);
+end
+
+numK = double(max(double(numK),double(max(gIX))));
+
+% sort traces by index
+im = M;
+[nLines,nFrames] = size(im);
+[~,I] = sort(gIX);
+im = im(I,:);
+
+%% convert imagesc effect into rgb matrix 'RGB'
+if dataFR, % is drawing cell-response fluorescence
+    cmap = gray(64);
+    im = AutoScaleImage0to1(im); % scaling to min/max 0/1
+    minlim = 0;
+    maxlim = 1;
+else % create blue-white-red map, for regression results
+    cmap = zeros(64,3);
+    cmap(:,1) = [linspace(0,1,32), linspace(1,1,32)];
+    cmap(:,2) = [linspace(0,1,32), linspace(1,0,32)];
+    cmap(:,3) = [linspace(1,1,32), linspace(1,0,32)];
+    minlim = -1; %min(min(im));
+    maxlim = 1; %max(max(im));
+end
+RGB = ImageToRGB(im,cmap,minlim,maxlim); % map image matrix to range of colormap
+
+%% add vertical color code
+if exist('clrmap','var'),
+    if strcmp(clrmap,'jet'),
+        temp = flipud(jet(numK));
+    else % 'hsv'
+        temp = hsv(round(numK*1.1));
+    end
+else % 'hsv'
+    temp = hsv(round(numK*1.1));
+end
+cmap2 = vertcat(temp(1:numK,:),[0,0,0]); % extend colormap to include black
+bwidth = max(round(nFrames/30),1);
+
+idx = gIX(I);
+ix_div = [find(diff(idx));length(idx)];
+
+bars = ones(nLines,bwidth);
+if numK>1,
+    bars(1:ix_div(1),:) = idx(ix_div(1));
+    for i = 2:length(ix_div),
+        % paint color
+        bars(ix_div(i-1)+1:ix_div(i),:) = idx(ix_div(i));
+    end
+end
+im_bars = reshape(cmap2(bars,:),[size(bars),3]);
+% add a white division margin
+div = ones(nLines,round(bwidth/2),3);
+% put 3 parts together
+im = horzcat(RGB,div,im_bars);
+
+%% plot figure
+
+[s1,s2,s3] = size(im);
+
+hold off;
+set(h1,'Position',[pos(1),pos(2)+pos(4)*barratio*3,pos(3),pos(4)*(1-4*barratio)]); 
+if s1<30,
+    temp = im;
+    im = ones(30,s2,s3);
+    im(1:s1,:,:) = temp;
+end
+image(im);
+set(gca, 'box', 'off')
+
+hold on;
+
+% plot cluster division lines
+plot([0.5,s2+0.5],[0.5,0.5],'k','Linewidth',0.5);
+if numK>1,
+    for i = 1:length(ix_div),% = numK-1,
+        y = ix_div(i)+0.5;
+        plot([0.5,s2+0.5],[y,y],'k','Linewidth',0.5);
+    end
+end
+
+ylabel(['Number of cells: ' num2str(numcell)]);
+set(gca,'YTick',[],'XTick',[]);
+set(gcf,'color',[1 1 1]);
+
+colormap gray;
+
+% write in number label
+x = s2*1.003;
+y_last = 0;
+for i = 1:length(ix_div),
+    % avoid label crowding
+    margin = 0.015*s1;
+    y0 = ix_div(i)+0.5;
+    y = max(y_last+margin,y0);
+    if i<length(ix_div),
+        ynext = ix_div(i+1);
+    else
+        ynext = y0+margin*2;
+    end
+    % draw if not squeezed too much away
+    if y < y0+margin*2 && y < 0.5*(y0+ynext), %nLines+stimheight-margin,
+        if exist('iswrite','var') && iswrite,
+            try
+                text(x,y,[num2str(idx(ix_div(i))) ': ' num2str(rankscore(i))],'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
+                    'FontUnits','normalized','FontSize',0.015,'Tag','label');
+            catch
+                disp('rankscore invalid');
+                text(x,y,num2str(idx(ix_div(i))),'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
+                    'FontUnits','normalized','FontSize',0.015,'Tag','label');
+            end
+            
+        else
+            text(x,y,num2str(idx(ix_div(i))),'HorizontalAlignment','Left','VerticalAlignment','Bottom',...
+                'FontUnits','normalized','FontSize',0.015,'Tag','label');
+        end
+        y_last = y;
+    end
+end
+
+%% Draw stim bars
+halfbarheight = 100;
+stimbar_ = GetStimBar(halfbarheight,stim);
+
+% pad with margin on right to fit combo-image size
+barlength = size(im,2);
+stimbar = ones(halfbarheight*2,barlength,3);
+stimbar(:,1:nFrames,:) = stimbar_(:,1:nFrames,:); % crop to size
+
+% top axis
+axes('Position',[pos(1),pos(2)+pos(4)*(1-barratio),pos(3),pos(4)*barratio]);
+image(stimbar);axis off;hold on;
+plot([1,length(stim)],[1,1],'k','Linewidth',0.5); % plot top border
+plot([0.5,0.5],[0,size(stimbar,1)],'k','Linewidth',0.5); % plot left/right border
+plot([length(stim),length(stim)],[0,size(stimbar,1)],'k','Linewidth',0.5);
+
+% bottom axis
+axes('Position',[pos(1),pos(2)+pos(4)*barratio*2,pos(3),pos(4)*barratio]);
+image(stimbar);axis off;hold on;
+plot([1,length(stim)],[size(stimbar,1),size(stimbar,1)],'k','Linewidth',0.5); % plot bottom border
+plot([0.5,0.5],[0,size(stimbar,1)],'k','Linewidth',0.5); % plot left/right border
+plot([length(stim),length(stim)],[0,size(stimbar,1)],'k','Linewidth',0.5);
+
+%% Draw fictive
+axes('Position',[pos(1),pos(2),pos(3),pos(4)*barratio*2]);
+
+temp = ones(size(fictive,1),barlength);
+temp(:,1:length(fictive)) = fictive;
+imagesc(temp);colormap hot
+set(gca,'YTick',[],'XTick',[]);
+set(gcf,'color',[1 1 1]);
+set(gca, 'box', 'off')
+hold on;axis ij;
+
+% plot division lines
+for i = 0:3,
+    y = i+0.5;
+    plot([0.5,length(fictive)+0.5],[y,y],'w','Linewidth',0.5);
+end
+% labels
+names = {'Left','Right','Forward','Raw L','Raw R'};
+x = -s2*0.05;
+for i = 1:5,
+    y = i;
+    text(x,y,names{i},'Fontsize',7);
+end
+
+%% horizontal axis: approximate time length, assuming 2 fps (real is 1.97 for now)
+set(gca,'XColor',[1 1 1]);
+if s2/2/60<5, % <5 min
+    l = xlabel(['Length of display: ~ ' num2str(s2/2) 'sec'],'Color',[0 0 0]);
+else
+    l = xlabel(['Length of display: ~ ' num2str(s2/2/60) 'min'],'Color',[0 0 0]);
+end
+
+%% Draw fish icon:
+if exist('fishpic.jpg','file')==2,
+    fishpic = imread('fishpic.jpg');
+    axes('Position',[0.015, 0.095, 0.03, 0.02]);
+    image(fishpic); axis off; axis equal
+end
+end
+
+%% Internal Functions
+
+function im = AutoScaleImage0to1(im)
+temp = mat2gray(im);
+low_high = stretchlim(temp,[0.005,0.995]);
+temp(temp<low_high(1)) = low_high(1);
+temp(temp>low_high(2)) = low_high(2);
+im = mat2gray(temp);
+end
+
+function RGB = ImageToRGB(im,cmap,minlim,maxlim)
+L = size(cmap,1);
+ix = round(interp1(linspace(minlim,maxlim,L),1:L,im,'linear','extrap'));
+RGB = reshape(cmap(ix,:),[size(ix) 3]); % Make RGB image from scaled.
+end
diff --git a/first version/DrawClustersOnMap_LSh.m b/first version/DrawClustersOnMap_LSh.m
new file mode 100644
index 0000000..5bbeb48
--- /dev/null
+++ b/first version/DrawClustersOnMap_LSh.m	
@@ -0,0 +1,137 @@
+function  [tot_image, dim_totimage] = DrawClustersOnMap_LSh(CInfo,cIX,gIX,numK,anat_yx,anat_yz,anat_zx,clrmap,full)
+%% formatting
+[s1,s2] = size(cIX);
+if s2>s1,
+    cIX = cIX';
+end
+[s1,s2] = size(gIX);
+if s2>s1,
+    gIX = gIX';
+end
+
+% down-sample
+if ~exist('full','var'),
+    displaymax = 8000;
+    if length(cIX) > displaymax,
+        skip = round(length(cIX)/displaymax);
+        cIX = cIX(1:skip:end,:);
+        gIX = gIX(1:skip:end,:);
+    end
+end
+
+% get numK
+if exist('numK','var'),
+    numK = double(max(numK,max(gIX)));
+else
+    numK = double(max(gIX));
+end
+
+if strcmp(clrmap,'jet'),
+    temp = flipud(jet(numK));
+else % 'hsv'
+    temp = hsv(round(numK*1.1));
+end
+cmap = temp(1:numK,:); % extend colormap to include black
+
+anat_YX = anat_yx/4;
+anat_YZ = anat_yz/4;
+anat_ZX = anat_zx/4;
+anat_ZX = flipud(anat_ZX);
+dimv_yx = size(anat_YX);
+dimv_yz = size(anat_YZ);
+dimv_zx = size(anat_ZX);
+
+k_zres = 20;
+anat_yz2=zeros(dimv_yz(1),dimv_yz(2)*k_zres,3);
+anat_zx2=zeros(dimv_zx(1)*k_zres,dimv_zx(2),3);
+dim_totimage = [dimv_yx(1)+dimv_zx(1)*k_zres+10,dimv_yx(2)+dimv_yz(2)*k_zres+10,3];
+tot_image=ones(dim_totimage);
+% tot_image=zeros(dim_totimage);
+% tot_image(:,dimv_yx(2)+(1:10),:)=1;
+
+% find index manipulation vector to darw circle
+circle=makeDisk2(7,15); % make mask of filled circle % (7,15)
+mask = zeros(dimv_yx(1),dimv_yx(2));
+mask(1:15,1:15) = circle;
+ix = find(mask);
+cix = sub2ind([dimv_yx(1),dimv_yx(2)],8,8);% 8
+circle_inds = ix - cix;
+
+yzplane_inds = -5:5;
+zxplane_inds = -5*dimv_zx(1):dimv_zx(1):5*dimv_zx(1);
+
+weight = 0.3-min(length(cIX)/1000/100,0.1);
+for j=1:length(cIX)
+    if ~isempty(CInfo(cIX(j)).center),
+        %% Y-X
+        %     cinds = sub2ind([dim_y,dim_x],cell_info(cellsIX(j)).center(1),cell_info(cellsIX(j)).center(2));
+        cinds=(CInfo(cIX(j)).center(2)-1)*dimv_yx(1)+CInfo(cIX(j)).center(1); % faster equivalent, lin px idx
+        labelinds=find((cinds+circle_inds)>0 & (cinds+circle_inds)<=dimv_yx(1)*dimv_yx(2)); % within bounds
+        ix = gIX(j);%find(U==gIX(j));U = 1:numK;
+        ixs = cinds+circle_inds(labelinds);
+        anat_YX(ixs) = cmap(ix,1)*weight + anat_YX(ixs)*(1-weight); % R
+        ixs = cinds+circle_inds(labelinds)+dimv_yx(1)*dimv_yx(2);
+        anat_YX(ixs) = cmap(ix,2)*weight + anat_YX(ixs)*(1-weight); % G
+        ixs = cinds+circle_inds(labelinds)+dimv_yx(1)*dimv_yx(2)*2;
+        anat_YX(ixs) = cmap(ix,3)*weight + anat_YX(ixs)*(1-weight); % B
+        
+        % Y-Z
+        zweight = weight/2;
+        %     cinds = sub2ind([dim_y,dim_z],cell_info(cellsIX(j)).center(1),cell_info(cellsIX(j)).slice);
+        cinds=(CInfo(cIX(j)).slice-1)*dimv_yz(1)+CInfo(cIX(j)).center(1);
+        labelinds=find((cinds+yzplane_inds)>0 & (cinds+yzplane_inds)<=dimv_yz(1)*dimv_yz(2));
+        ixs = cinds+yzplane_inds(labelinds);
+        anat_YZ(ixs) = cmap(ix,1)*zweight + anat_YZ(ixs)*(1-zweight);
+        ixs = cinds+yzplane_inds(labelinds)+dimv_yz(1)*dimv_yz(2);
+        anat_YZ(ixs) = cmap(ix,2)*zweight + anat_YZ(ixs)*(1-zweight);
+        ixs = cinds+yzplane_inds(labelinds)+dimv_yz(1)*dimv_yz(2)*2;
+        anat_YZ(ixs) = cmap(ix,3)*zweight + anat_YZ(ixs)*(1-zweight);
+                
+        % Z-X
+        zweight = weight/2;
+        %     cinds = sub2ind([dim_y,dim_z],cell_info(cellsIX(j)).center(1),cell_info(cellsIX(j)).slice);
+        cinds=(CInfo(cIX(j)).center(2)-1)*dimv_zx(1) +(CInfo(cIX(j)).slice);
+        labelinds=find((cinds+zxplane_inds)>0 & (cinds+zxplane_inds)<=dimv_zx(1)*dimv_zx(2));
+        ixs = cinds+zxplane_inds(labelinds);
+        anat_ZX(ixs) = cmap(ix,1)*zweight + anat_ZX(ixs)*(1-zweight);
+        ixs = cinds+zxplane_inds(labelinds)+dimv_zx(1)*dimv_zx(2);
+        anat_ZX(ixs) = cmap(ix,2)*zweight + anat_ZX(ixs)*(1-zweight);
+        ixs = cinds+zxplane_inds(labelinds)+dimv_zx(1)*dimv_zx(2)*2;
+        anat_ZX(ixs) = cmap(ix,3)*zweight + anat_ZX(ixs)*(1-zweight);
+    end
+end
+% rescale (low-res) z dimension
+for k=1:3
+    anat_yz2(:,:,1) = imresize(anat_YZ(:,:,1), [dimv_yz(1), dimv_yz(2)*k_zres],'nearest');
+    anat_yz2(:,:,2) = imresize(anat_YZ(:,:,2), [dimv_yz(1), dimv_yz(2)*k_zres],'nearest');
+    anat_yz2(:,:,3) = imresize(anat_YZ(:,:,3), [dimv_yz(1), dimv_yz(2)*k_zres],'nearest');
+    
+    anat_zx2(:,:,1) = imresize(anat_ZX(:,:,1), [dimv_zx(1)*k_zres, dimv_zx(2)],'nearest');
+    anat_zx2(:,:,2) = imresize(anat_ZX(:,:,2), [dimv_zx(1)*k_zres, dimv_zx(2)],'nearest');
+    anat_zx2(:,:,3) = imresize(anat_ZX(:,:,3), [dimv_zx(1)*k_zres, dimv_zx(2)],'nearest');
+end
+
+tot_image(dimv_zx(1)*k_zres+11:end,1:dimv_yx(2),:) = anat_YX;
+tot_image(dimv_zx(1)*k_zres+11:end,dimv_yx(2)+11:end,:) = anat_yz2;
+tot_image(1:dimv_zx(1)*k_zres,1:dimv_zx(2),:) = flipud(anat_zx2);
+
+tot_image(tot_image(:)>1) = 1;
+tot_image(tot_image(:)<0) = 0;
+
+image(tot_image);
+axis image;axis off
+
+end
+
+function out = makeDisk2(radius, dim)
+center=floor(dim/2)+1;
+out=zeros(dim);
+for x=1:dim
+    for y=1:dim
+        if norm([x,y]-[center,center])<=radius
+            out(x,y)=1;
+        end
+    end
+end
+
+end
\ No newline at end of file
diff --git a/FFT_test.m b/first version/FFT_test.m
similarity index 100%
rename from FFT_test.m
rename to first version/FFT_test.m
diff --git a/first version/GUI_FishExplorer.m b/first version/GUI_FishExplorer.m
new file mode 100644
index 0000000..a417875
--- /dev/null
+++ b/first version/GUI_FishExplorer.m	
@@ -0,0 +1,3446 @@
+%%% Interactive app for exploratory analysis of calcium imaging data 
+% (with stimulus, behavior, and anatomy)
+
+% Input calcium data: 1 trace per cell/ROI, ~50,000 cells per fish
+% load collection of cells from multiple fish, or load full data of single fish individually
+% main outputs: GUI plots, clusters saved into .mat, export variables to MATLAB workspace
+
+% Tip: to see the structure of this code, use 'Ctrl' + 'm' + '=' to collapse all cells.
+% UI controls are organized by tabs and then by rows, instructions and 
+% comments are where they are constructed ('User Interface:' -> function hfig... ->)
+% General internal functions are at the end, some specialized .m functions are outside.
+
+% Written in Matlab R2014b running on Windows 7.
+
+% - Xiuye Chen (xiuyechen@gmail.com), Engert Lab, Spring 2015
+
+
+% To start, run the following in separate script or command window:
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+% clear all;close all;clc
+% global VAR;
+% load('VAR_current.mat','VAR');
+% load('CONSTs_current.mat','CONSTs');
+% hfig = GUI_FishExplorer(CONSTs,VAR);
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+% Full datasets are stored as 'CONST_F?.mat' (? = fish number), to load from the GUI
+
+%% User Interface:
+% function hfig = GUI_FishExplorer(CONSTs,VAR,CONST,i_fish)
+function [hfig,fcns] = GUI_FishExplorer(data_dir,CONSTs,VAR,flag_script,var_script)
+if exist('flag_script','var')
+    if ~exist('var_script','var')
+        var_script={};
+    end
+    runscript(flag_script,var_script);
+    return
+end
+
+%% Make figure
+scrn = get(0,'Screensize');
+hfig = figure('Position',[scrn(3)*0.1 scrn(4)*0.05 scrn(3)*0.85 scrn(4)*0.86],...% [50 100 1700 900]
+    'Name','GUI_LSh','DeleteFcn',@closefigure_Callback);
+hold off; axis off
+
+%% Folder setup
+% directory for full fish data (.mat)
+setappdata(hfig,'data_dir',data_dir);
+
+% copy of VAR files will be saved into this subfolder:
+currentfolder = pwd;
+arcmatfolder = [currentfolder '\arc mat'];
+if ~exist(arcmatfolder, 'dir')
+  mkdir(arcmatfolder);
+end
+setappdata(hfig,'arcmatfolder',arcmatfolder);
+
+%% Pass external variables into appdata (stored with main figure handle)
+setappdata(hfig,'CONSTs',CONSTs);
+setappdata(hfig,'VAR',VAR);
+nFish = length(CONSTs);
+
+% parameters / constants
+setappdata(hfig,'z_res',19.7); % resoltion in z, um per slice
+% fpsec = 1.97; % hard-coded in ext function 'GetStimRegressor.m'
+% approx fpsec of 2 used in ext function 'DrawCluster.m'
+
+% cache 
+bC = []; % Cache for going b-ack (bad abbr)
+fC = []; % Cache for going f-orward
+bC.cIX = cell(1,1);
+bC.gIX = cell(1,1);
+bC.numK = cell(1,1);
+fC.cIX = cell(1,1);
+fC.gIX = cell(1,1);
+fC.numK = cell(1,1);
+setappdata(hfig,'bCache',bC);
+setappdata(hfig,'fCache',fC);
+
+% initialization
+i_fish = 1;
+QuickUpdateFish(hfig,i_fish,'init');
+
+%% Initialize internal params into appdata
+% fish protocol sets (different sets have different parameters)
+M_fish_set = [1, 1, 1, 1, 1, 1, 1, 2, 1]; % M = Matrix
+setappdata(hfig,'M_fish_set',M_fish_set);
+
+% thresholds
+thres_merge = 0.9;
+thres_split = 0.7;
+thres_reg = 0.7;
+thres_size = 10;
+setappdata(hfig,'thres_merge',thres_merge);
+setappdata(hfig,'thres_split',thres_split); % for function 'pushbutton_iter_split'
+setappdata(hfig,'thres_reg',thres_reg); % regression threshold, ~correlation coeff
+setappdata(hfig,'thres_size',thres_size); % min size for clusters
+
+% variables 
+% (not sure all these need to be initialized, probably not complete either)
+setappdata(hfig,'clrmap','hsv');
+setappdata(hfig,'opID',0);
+setappdata(hfig,'rankID',0); 
+setappdata(hfig,'rankscore',[]);
+setappdata(hfig,'isCentroid',0);
+setappdata(hfig,'isWkmeans',1); % in autoclustering, with/without kmeans
+setappdata(hfig,'isflipstim',0); % flag for flip updown
+setappdata(hfig,'regchoice',{1,1}); % regressor choice; first cell,1=stim,2=motor,3=centroid
+setappdata(hfig,'isfullfish',0); % no if QuickUpdateFish, yes if LoadFullFish
+setappdata(hfig,'isPlotCorrHist',0); % option for regression
+setappdata(hfig,'dataFR',1); % left plot data type, 1 for fluo trace and 0 for regression result
+setappdata(hfig,'isPlotReg',1); % plot regressor when selecting it
+setappdata(hfig,'hierinplace',1); % hier. partitioning, no reordering
+
+
+Class = getappdata(hfig,'Class');
+classID = numel(Class);
+setappdata(hfig,'classID',classID);
+setappdata(hfig,'classheader','Test:');
+setappdata(hfig,'newclassname','');
+
+Cluster = getappdata(hfig,'Cluster');
+clusID = 1; %numel(Cluster);
+setappdata(hfig,'clusID',clusID);
+setappdata(hfig,'clusheader','Test:');
+setappdata(hfig,'newclusname','');
+
+setappdata(hfig,'cIX',Cluster(clusID).cIX); % cell-IndeX
+setappdata(hfig,'gIX',Cluster(clusID).gIX); % group-IndeX
+
+%% Create UI controls
+set(gcf,'DefaultUicontrolUnits','normalized');
+set(gcf,'defaultUicontrolBackgroundColor',[1 1 1]);
+
+% tab group setup
+tgroup = uitabgroup('Parent', hfig, 'Position', [0.05,0.88,0.91,0.12]);
+numtabs = 5;
+tab = cell(1,numtabs);
+M_names = {'General','Operations','Regression','Clustering etc.','Saved Clusters'};
+for i = 1:numtabs,
+    tab{i} = uitab('Parent', tgroup, 'BackgroundColor', [1,1,1], 'Title', M_names{i});
+end
+
+% grid setup, to help align display elements
+rheight = 0.2;
+yrow = 0.7:-0.33:0;%0.97:-0.03:0.88;
+bwidth = 0.03;
+grid = 0:bwidth+0.001:1;
+
+% various UI element handles ('global' easier than passing around..)
+global hback hfwd hclassname hclassmenu hclusgroupmenu hclusmenu hclusname...
+    hdatamenu hopID hcentroid hdataFR hloadfish hfishnum hstimreg hmotorreg...
+    hcentroidreg;
+
+%% UI ----- tab one ----- (General)
+i_tab = 1;
+
+%% UI row 1: File 
+i_row = 1;
+i = 1;n = 0;
+
+i=i+n;
+n=2; % saves 'VAR' to workspace. 'VAR' contains clustering indices (Class and Clusgroup) for all fish
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Quick save',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_save_Callback);
+
+i=i+n;
+n=2; % saves 'VAR' both to workspace and to 'VAR_current.mat' and to arc folder
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Save .mat',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_savemat_Callback);
+
+i=i+n;
+n=2; % plots selected cells on anatomy z-stack, display and save tiff stack in current directory
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Save Zstack',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_writeZstack_Callback);
+
+i=i+n;
+n=2; % make new figure without the GUI components, can save manually from default menu
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Popup plot',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_popupplot_Callback);
+
+i=i+n;
+n=3; % export main working variables to workspace, can customize!
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Export to workspace',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_exporttoworkspace_Callback);
+
+%% UI row 2: Load 
+i_row = 2;
+i = 1;n = 0;
+
+i=i+n;
+n=2; % this design is underused now... Quick-load only depends on CONSTs,
+% which is a minimum collection of clusters from all fish, so you can load
+% the program without full single-fish data. eventually can use this  
+% platform to do things across fish (like after anatomical alignment).
+uicontrol('Parent',tab{i_tab},'Style','text','String','Quick-load fish:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1; % loads 'CONSTs_current.mat' from current directory
+temp = {}; for j = 1:nFish, temp = [temp,{num2str(j)}];end
+hfishnum = uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',temp,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@popup_fishmenu_Callback);
+
+i=i+n;
+n=2; % loads full single-fish data from CONST_F?.mat
+uicontrol('Parent',tab{i_tab},'Style','text','String','Load full fish:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=2; % these CONST_F?.mat are now saved as uncompressed version 6 .mat, loads faster
+temp = {}; for j = 1:nFish, temp = [temp,{num2str(j)}];end
+temp = [{'(choose)'},temp];
+hloadfish = uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',temp,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@popup_loadfullfishmenu_Callback);
+
+i=i+n;
+n=2; % load different presentations (average, all reps etc), pre-stored in CONST(s)
+uicontrol('Parent',tab{i_tab},'Style','text','String','Data type:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=3; 
+hdatamenu = uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',CONSTs{i_fish}.datanames,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@popup_datamenu_Callback);
+
+i=i+n;
+n=4; % only centroids (~mean) of clusters shown on left-side plot, the rest is unchanged
+hcentroid = uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Display centroids of clusters',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@checkbox_showcentroids_Callback);
+
+%% UI ----- tab two ----- (Operations)
+i_tab = 2;
+
+%% UI row 1: Range
+i_row = 1;
+i = 1;n = 0;
+
+i=i+n;
+n=2; % saves up to 20 steps backwards (datatype/datamenu change does not count)
+hback = uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Back',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_back_Callback);
+
+i=i+n;
+n=2; % same, 20 steps forward if applicable
+hfwd = uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Forward',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_forward_Callback);
+
+i=i+n;
+n=3; % Choose range of clusters to keep. format: e.g. '1:2,4-6,8:end'
+uicontrol('Parent',tab{i_tab},'Style','text','String','Choose cluster range:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1; 
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_choose_range_Callback);
+
+i=i+n;
+n=2; % Choose range of clusters to exclude. format: e.g. '1:2,4-6,8:end'
+uicontrol('Parent',tab{i_tab},'Style','text','String','Exclude:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_exclude_range_Callback);
+
+i=i+n;
+n=1; % Choose range of clusters to fuse/combine into single cluster. format: e.g. '1:2,4-6,8:end'
+uicontrol('Parent',tab{i_tab},'Style','text','String','Fuse:',... % (eg 1:2,3-5)
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_fuse_range_Callback);
+
+i=i+n;
+n=2; % can choose range in time, but can't see time axes to decide... resets when choosing datamenu
+uicontrol('Parent',tab{i_tab},'Style','text','String','Time-range:',... % (eg 1:2,3-5)
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_t_range_Callback);
+
+%% UI row 2: Operations
+i_row = 2;
+i = 1;n = 0;
+
+i=i+n;
+n=2; % operates between the current cell selection and the next (in this order). 
+uicontrol('Parent',tab{i_tab},'Style','text','String','Set operations:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n; % 'setdiff' is current minus next, 'rev setdiff' is next minus current. 
+n=2; % smartUnion = SmartUnique, cells belonging to 2 clusters goes to the more correlated one
+menu = {'(choose)','union','intersect','setdiff','rev setdiff','setxor','smartUnion'};
+hopID = uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',menu,'Value',1,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@popup_operations_Callback});
+
+i=i+n;
+n=2; % rank clusters based on various criteria (to choose)
+uicontrol('Parent',tab{i_tab},'Style','text','String','Rank by:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n; % 'hier' is the same as default (used after every k-means);'stim-lock' uses std across reps;
+n=2; % motor stuff uses the best alignment (by cross-correlation) with the fictive trace;
+% L+R is average of L & R; stim-motor is combines 'stim-lock' w 'motor' with arbituary weighting.
+menu = {'(choose)','hier.','size','stim-lock','corr',...
+    'motor','L motor','R motor','L+R motor','stim-motor'}; 
+uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',menu,'Value',1,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@popup_ranking_Callback});
+
+i=i+n;
+n=3; % cluster indices will rank from 1 to number of clusters
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Sqeeze clusters',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_sqeeze_Callback);
+
+i=i+n;
+n=3; % flip the sequenc of clusters, first becomes last
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Flip up-down',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_flipud_Callback);
+
+i=i+n;
+n=3; % switch between 2 colormaps now, jet and a cropped version of hsv (so not all circular)
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Switch colormap',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_clrmap_Callback});
+
+%% UI row 3: Anatomy
+i_row = 3;
+i = 1;n = 0;
+
+i=i+n;
+n=4; % Draw a polygon on anatomy maps to select the cells within those boundaries
+uicontrol('Parent',tab{i_tab},'Style','text','String','Draw on anatomy map to crop:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n; % Draw on the yx-view (main view) 
+n=2; % Click to make new vertex, double click to connect to first vertex, 
+% then optionally drag vertices to reposition, and finally double click again to set
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Draw yx',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_polygon_yx_Callback);
+
+i=i+n;
+n=2; % Draw on yz-view (side projection), same as above
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Draw yz',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_polygon_yz_Callback);
+
+i=i+n;
+n=2; % Draw on zx-view (front projection), same as above
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Draw zx',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_polygon_zx_Callback);
+
+i=i+n;
+n=5; 
+uicontrol('Parent',tab{i_tab},'Style','text','String','Select all cells within boundaries:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=2; % selects ALL cells contained in dataset that are within the convex shape defined by current cells
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Convex hull',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_withinConvexHull_Callback);
+
+%% UI ----- tab three ----- (Regression)
+i_tab = 3;
+
+%% UI row 1: regressor
+i_row = 1; % Step 1:
+i = 1;n = 0; % Choose one type of regressor here, choice highlighted in yellow
+
+i=i+n;
+n=2; % stimulus regressors, go to 'GetStimRegressor.m' to add/update
+uicontrol('Parent',tab{i_tab},'Style','text','String','Stim reg.:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=3; % (updated when loading fish)
+menu = {'(choose)',''};
+hstimreg = uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',menu,'Value',1,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@popup_getstimreg_Callback);
+
+i=i+n; % motor regressors from fictive, not yet convolved/adjusted for time lag
+n=2; % go to 'GetMotorRegressor.m' to add/update
+uicontrol('Parent',tab{i_tab},'Style','text','String','Motor reg.:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=2; % (unlike stim regressors, names hardcoded, not importet from regressor...)
+menu = {'(choose)','right swims','left swims','forward swims','raw right','raw left','raw average'};
+hmotorreg = uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',menu,'Value',1,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@popup_getmotorreg_Callback);
+
+i=i+n;
+n=2; % if checked, plot regressor during selection, together with stim and motor
+uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Plot regressor',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],'Value',1,...
+    'Callback',@checkbox_popupplotreg_Callback);
+
+i=i+n+1;
+n=4; % Get centroid (~mean) of selected cluster as regressor
+uicontrol('Parent',tab{i_tab},'Style','text','String','Regressor from centroid #:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+hcentroidreg = uicontrol('Parent',tab{i_tab},'Style','edit','String',num2str(1),...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_ctrdID_as_reg_Callback);
+
+i=i+n;
+n=2; % too complicated... chomp up centroid, then order as if stimulus were left/right inverted
+uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Flip stim',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@checkbox_flipstim_Callback);
+
+%% UI row 2: regression
+i_row = 2; % Step 2:
+i = 1;n = 0; % Choose regression, using the regressor chosen above, search in full dataset
+
+i=i+n;
+n=3;
+uicontrol('Parent',tab{i_tab},'Style','text','String','Choose regression ->',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=2; % do regression, show all cells with correlation coeff (with regressor) above threshold
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Corr. threshold:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_thres_regression_Callback);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit','String',num2str(thres_reg),...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_regthres_Callback);
+
+i=i+n;
+n=2; % optionally plot histogram of correlation values for all cells in dataset, visualize cut-off
+hdataFR = uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Plot corr. hist',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@checkbox_plotcorrhist_Callback);
+
+i=i+n;
+n=3; % probably not so useful. Show top n cells of highest correlation coeff with regressor
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','top corr., number limit:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_topnum_regression_Callback});
+
+i=i+n;
+n=1; % specify n for above
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_topCorrNumber_Callback);
+
+i=i+n+1; % more automatic, do a regression with every centroid, then combine (with 'SmartUnique'
+n=4; % i.e. overlapping cells are assigned to the cluster with which the correlation is the highest)
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Regression with all centroids',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_AllCentroidRegression_Callback});
+
+i=i+n; % this is a remnant button from a failed experiment, idea was to iterate the regression process
+n=2; % until the cluster converges, but most of the time it doesn't...
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','iter.reg','Enable','off',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_IterCentroidRegression_Callback});
+
+%% UI row 3: (TBD)
+i_row = 3; % old idea: display correlation values of multiple regressors (instead of fluo. trace)
+i = 1;n = 0; % and then can cluster that and further manipulate... 
+
+i=i+n;
+n=4; % did not implement combination of regressors in this version... but display option is coded (dataFR==0)
+uicontrol('Parent',tab{i_tab},'Style','text','String','regressor combos??(TBD)',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=4; % also data storage is coded, 'M' could be loaded from M_0_fluo or M_0_reg (storing reg corr values)
+hdataFR = uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Display fluo./regression',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@checkbox_dataFR_Callback);
+
+%% UI ----- tab four ----- (Clustering etc.)
+i_tab = 4;
+
+%% UI row 1: k-means
+i_row = 1;
+i = 1;n = 0;
+
+i=i+n;
+n=2; % k-means clustering
+uicontrol('Parent',tab{i_tab},'Style','text','String','k-means, k =',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_kmeans_Callback);
+
+i=i+n;
+n=4; % anatomy is added to the fluo trace as new dimensions, and (arbituarily) weighted strongly
+uicontrol('Parent',tab{i_tab},'Style','text','String','k-means with anatomy, k =',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_kmeans2_Callback);
+
+i=i+n; % trying to use Silhouette to evaluate cluster quality, find peak to determine optimal k,
+n=3; % then display results with that k. But have not set k-means to replicate (speed concern), can be very noisy
+uicontrol('Parent',tab{i_tab},'Style','text','String','Find best k in range:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight])
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_kmeans_elbow_Callback);
+
+%% UI row 2: Auto-clustering
+i_row = 2;
+i = 1;n = 0;
+
+i=i+n; % Adjacent clusters (arranged by hier.) will be merged
+n=4;  % if correlation between centroids is above merging threshold
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Merge thres. (corr. based)',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_merge_Callback});
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit','String',num2str(thres_merge),...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@edit_mergethres_Callback});
+
+i=i+n; % further split initial clusters so that the average within-cluster corr coeff is above thres
+n=2; % (not so iterative anymore, but could easily restore)
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Iter. split',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_iter_split});
+
+i=i+n;
+n=1; 
+uicontrol('Parent',tab{i_tab},'Style','edit','String',num2str(thres_split),...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@edit_splitthres_Callback});
+
+i=i+n;
+n=3; % minimal size of cluster, otherwise delete at the end
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Cluster size thres.',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_thressize_Callback});
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit','String',num2str(thres_size),...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@edit_sizethres_Callback});
+
+i=i+n+1; % longest script here. Splits clusters and prunes them, to yield only very tight clusters.
+n=3; % really pretty results, but takes a while when regressing with every centroid. Read code for details.
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Full Auto-Clustering',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_autoclus_Callback});
+
+i=i+n;
+n=4; % by default it starts with a k-mean of 20 of the current cells. Could skip that if already clustered.
+uicontrol('Parent',tab{i_tab},'Style','checkbox','String','(starting with k-mean)',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],'Value',1,...
+    'Callback',@checkbox_wkmeans_Callback);
+
+%% UI row 3: misc plots
+i_row = 3;
+i = 1;n = 0;
+
+i=i+n; % k-means are always ranked like in hierachical clustering ~ optimal leaf order
+n=2; % here you can rank them again and plot the dengrogram.
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Hier. plot',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_hierplot_Callback);
+
+i=i+n; % partitioning based on hier. clustering
+n=2; % choose between max cluster numbers...
+uicontrol('Parent',tab{i_tab},'Style','text','String','Hier.cut, max n:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_hierpartn_Callback);
+
+i=i+n; % partitioning based on hier. clustering
+n=3; % ...or set correlation value threshold
+uicontrol('Parent',tab{i_tab},'Style','text','String','Hier.cut, corr thres:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_hierpartthres_Callback);
+
+i=i+n; % hier. partition in place, i.e. without rearranging order clusters
+n=3; 
+uicontrol('Parent',tab{i_tab},'Style','checkbox','String','Hier.cut in place',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],'Value',1,...
+    'Callback',@checkbox_hierinplace_Callback);
+
+i=i+n;
+n=2; % Plots the correlation between all current clusters as a matrix
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Corr. plot',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_corrplot_Callback});
+
+i=i+n;
+n=2; % random plots not really worth consolidating, only for 1-click convenience
+uicontrol('Parent',tab{i_tab},'Style','text','String','temp. plots:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight])
+
+i=i+n;
+n=3; % e.g. this one looks at the history for the fish presented with 4 B/W stimuli
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','plot 4x4(stim)',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_plot4x4_Callback});
+
+%% UI ----- tab five ----- (Saved Clusters)
+i_tab = 5;
+
+%% UI row 1: Class
+i_row = 1; % There is not really a difference between Class and Cluster
+i = 1;n = 0; % I just like to have 2 layers, so I save bigger sets as Class, smaller things as Cluster
+% also now Clusters come in cluster-groups ('Clusgroup'), i.e. multiple groups of Clusters...
+% the saving is sort of a pain though, several Update___ internal functions dedicated to this.
+
+i=i+n;
+n=1; % 
+uicontrol('Parent',tab{i_tab},'Style','text','String','Class:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=4;
+temp = MakeMenu({Class.name});
+hclassmenu = uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',temp,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@popup_classmenu_Callback);
+
+i=i+n;
+n=1; % just edit and press enter to edit current name
+uicontrol('Parent',tab{i_tab},'Style','text','String','Edit:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=3;
+hclassname = uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@edit_editclassname_Callback});
+set(hclassname,'String',Class(classID).name);
+
+i=i+n;
+n=2; % save the current cells, with current cluster divisions, as 1 class
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Save class',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_saveclass_Callback);
+
+i=i+n;
+n=1; % not really useful, just to help me organize the Class name
+uicontrol('Parent',tab{i_tab},'Style','text','String','Header:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1; 
+uicontrol('Parent',tab{i_tab},'Style','edit','String','Test',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_class_header_Callback);
+
+i=i+n;
+n=1; % write down new name for new cluster
+uicontrol('Parent',tab{i_tab},'Style','text','String','New:',... 
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=2;
+uicontrol('Parent',tab{i_tab},'Style','edit','String','(blank)',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_newclassname_Callback);
+
+i=i+n;
+n=2; % new class is made when pressing this button
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Make class',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_makeclass_Callback);
+
+i=i+n;
+n=2; % will prompt you to confirm
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Delete Class!',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_delclass_Callback});
+
+%% UI row 2: Cluster
+i_row = 2; % Virtually the same as Class, except added a ranking option
+i = 1;n = 0;
+
+i=i+n;
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','text','String','Cluster (group):',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+num = length(VAR(i_fish).ClusGroup);
+temp = {}; for j = 1:num, temp = [temp,{num2str(j)}];end
+hclusgroupmenu = uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',temp,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@popup_clusgroupmenu_Callback);
+
+i=i+n;
+n=3;
+temp = MakeMenu({Cluster.name});
+hclusmenu = uicontrol('Parent',tab{i_tab},'Style','popupmenu','String',temp,...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@popup_clusmenu_Callback);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','text','String','Edit:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=3;
+hclusname = uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@edit_editclusname_Callback});
+set(hclusname,'String',Cluster(clusID).name);
+
+i=i+n;
+n=2;
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Save cluster',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_saveclus_Callback);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','text','String','Header:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit','String','Test',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_clus_header_Callback);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','text','String','New:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=2;
+uicontrol('Parent',tab{i_tab},'Style','edit','String','(blank)',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_newclusname_Callback);
+
+i=i+n;
+n=2;
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Make cluster',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@pushbutton_makeclus_Callback);
+
+i=i+n;
+n=2;
+uicontrol('Parent',tab{i_tab},'Style','text','String','Set rank:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@edit_setrank_Callback});
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','text','String','Notes:',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=2;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@edit_notes_Callback});
+
+i=i+n;
+n=2;
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','Delete Cluster!',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_delclus_Callback});
+
+%% UI row 3: misc
+i_row = 3; % continuation, dealing with the Cluster groups 
+i = 1;n = 0; % (Cluster-group number: number menu before the Cluster-name menu)
+
+i=i+n;
+n=2; % Combines the chosen clusters into one view (can save as 1 class or cluster then)
+uicontrol('Parent',tab{i_tab},'Style','text','String','Union(cluster):',... % (eg 1,3-5)
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight]);
+
+i=i+n;
+n=1;
+uicontrol('Parent',tab{i_tab},'Style','edit',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',@edit_clusUnion_Callback);
+
+i=i+n; % just adds a new number to the Clustergroup-number menu, 
+n=3; % and saves current view as the first cluster in the new Clustergroup
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','new Clustergroup',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_newclusgroup_Callback});
+
+i=i+n;
+n=3; % delete current
+uicontrol('Parent',tab{i_tab},'Style','pushbutton','String','del Clustergroup',...
+    'Position',[grid(i) yrow(i_row) bwidth*n rheight],...
+    'Callback',{@pushbutton_delclusgroup_Callback});
+
+%% Load figure
+
+UpdateClusID(hfig,clusID);
+
+%% get local function handles
+
+fcns = localfunctions;
+
+end
+
+%% Callback functions for UI elements: 
+
+%% ----- tab one ----- (General)
+
+%% row 1: File
+
+function pushbutton_save_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+i_fish = getappdata(hfig,'i_fish');
+global VAR;
+VAR(i_fish).Class = getappdata(hfig,'Class');
+VAR(i_fish).ClusGroup = CurrentClusGroup(hfig);
+disp('saved to workspace ''VAR''');
+end
+
+function pushbutton_savemat_Callback(hObject,~)
+disp('saving...');
+hfig = getParentFigure(hObject);
+i_fish = getappdata(hfig,'i_fish');
+arcmatfolder = getappdata(hfig,'arcmatfolder');
+global VAR;
+VAR(i_fish).Class = getappdata(hfig,'Class');
+VAR(i_fish).ClusGroup = CurrentClusGroup(hfig);
+timestamp  = datestr(now,'mmddyy_HHMM');
+matname = [timestamp '.mat'];
+save(fullfile(arcmatfolder,matname),'VAR','-v6');
+save('VAR_current.mat','VAR','-v6');
+disp('saved both to workspace and .mat');
+end
+
+function pushbutton_writeZstack_Callback(hObject,~) 
+disp('preparing z-stack...');
+isMarkAllCells = 1;
+
+hfig = getParentFigure(hObject);
+isfullfish = getappdata(hfig,'isfullfish');
+if ~isfullfish,
+    errordlg('Load full fish first!');
+    return;
+end
+
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+CInfo = getappdata(hfig,'CInfo');
+ave_stack = getappdata(hfig,'ave_stack');
+
+timestamp = datestr(now,'mmddyy_HHMMSS');
+tiffName = ['stack_' timestamp '.tif'];
+
+U = unique(gIX);
+numK = length(U);
+
+% left half: stack with cells marked; 
+% right half: original anatomy, or mark all cells
+
+ave_stack2=zeros(size(ave_stack,1), size(ave_stack,2)*2, size(ave_stack,3) ,3);
+nPlanes=size(ave_stack,3);
+dimv_yxz=size(ave_stack);
+stacklen=numel(ave_stack);
+
+circle=makeDisk2(10,21);
+[r, v]=find(circle);
+r=r-11;v=v-11;
+circle_inds  = r*dimv_yxz(1)+v;
+cmap = hsv(numK);
+weight = 0.3;
+
+for i=1:nPlanes,
+    ave_stack2(:,:,i,:)=repmat(imNormalize99(ave_stack(:,:,i)),[1 2 1 3]);
+end
+
+for j=1:length(cIX)
+    cinds=(CInfo(cIX(j)).center(2)-1)*dimv_yxz(1)+CInfo(cIX(j)).center(1);
+    labelinds=find((cinds+circle_inds)>0 & (cinds+circle_inds)<=dimv_yxz(1)*dimv_yxz(2));
+    zinds=dimv_yxz(1)*dimv_yxz(2)*2*(CInfo(cIX(j)).slice-1);
+    ix = find(U==gIX(j));
+    ixs = cinds+circle_inds(labelinds)+zinds;
+    ave_stack2(ixs)=cmap(ix,1)*weight + ave_stack2(ixs)*(1-weight);
+    ixs = cinds+circle_inds(labelinds)+zinds+stacklen*2;
+    ave_stack2(ixs)=cmap(ix,2)*weight + ave_stack2(ixs)*(1-weight);
+    ixs = cinds+circle_inds(labelinds)+zinds+stacklen*4;
+    ave_stack2(ixs)=cmap(ix,3)*weight + ave_stack2(ixs)*(1-weight);
+end
+if isMarkAllCells,
+    clr = [0,1,0];
+    numcell = getappdata(hfig,'numcell');
+    for j=1:numcell,
+        shift = dimv_yxz(1)*dimv_yxz(2);
+        cinds=(CInfo(j).center(2)-1)*dimv_yxz(1)+CInfo(j).center(1);
+        labelinds=find((cinds+circle_inds)>0 & (cinds+circle_inds)<=dimv_yxz(1)*dimv_yxz(2));
+        zinds=dimv_yxz(1)*dimv_yxz(2)*2*(CInfo(j).slice-1);
+        ixs = cinds+circle_inds(labelinds)+zinds + shift;
+        ave_stack2(ixs)=clr(1)*weight + ave_stack2(ixs)*(1-weight);
+        ixs = cinds+circle_inds(labelinds)+zinds+stacklen*2 + shift;
+        ave_stack2(ixs)=clr(2)*weight + ave_stack2(ixs)*(1-weight);
+        ixs = cinds+circle_inds(labelinds)+zinds+stacklen*4 + shift;
+        ave_stack2(ixs)=clr(3)*weight + ave_stack2(ixs)*(1-weight);
+    end
+end
+h = figure;
+for i_plane = 1:nPlanes,
+    im = squeeze(ave_stack2(:,:,i_plane,:));
+    image(im);
+    % save tiff
+    if (i_plane == 1)
+        imwrite(im, tiffName, 'compression','none','writemode','overwrite')
+    else
+        imwrite(im, tiffName, 'compression','none','writemode','append')
+    end
+    pause(0.2)
+end
+close(h)
+end
+
+function out = makeDisk2(radius, dim)
+center=floor(dim/2)+1;
+out=zeros(dim);
+for x=1:dim
+    for y=1:dim
+        if norm([x,y]-[center,center])<=radius
+            out(x,y)=1;
+        end
+    end
+end
+end
+
+function pushbutton_popupplot_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+
+% same as function RefreshFigure(hfig), except new axes not global
+
+% figure('Position',[50,100,853,512]); % right plot size: 2048x1706
+% h1 = axes('Position',[0, 0, 0.5, 1]); % left ~subplot
+% h2 = axes('Position',[0.50, 0, 0.5, 1]); % right ~subplot
+figure('Position',[50,100,1600,900]);
+h1 = axes('Position',[0.05, 0.03, 0.53, 0.94]); % left ~subplot
+h2 = axes('Position',[0.61, 0.03, 0.35, 0.94]); % right ~subplot
+
+CInfo = getappdata(hfig,'CInfo');
+M = getappdata(hfig,'M');
+dataFR = getappdata(hfig,'dataFR');
+fictive = getappdata(hfig,'fictive');
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+
+numK = getappdata(hfig,'numK');
+stim = getappdata(hfig,'stim');
+anat_yx = getappdata(hfig,'anat_yx');
+anat_yz = getappdata(hfig,'anat_yz');
+anat_zx = getappdata(hfig,'anat_zx');
+isCentroid = getappdata(hfig,'isCentroid');
+clrmap = getappdata(hfig,'clrmap');
+rankscore = getappdata(hfig,'rankscore');
+rankID = getappdata(hfig,'rankID');
+
+iswrite = (rankID>=2);
+
+% left subplot
+axes(h1);
+if isCentroid,
+    [C,~] = FindCentroid(gIX,M);
+    DrawClusters(h1,C,unique(gIX),dataFR,numK,stim,fictive,clrmap,rankscore,iswrite);
+else
+    DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite);
+end
+
+% right subplot
+axes(h2);
+DrawClustersOnMap_LSh(CInfo,cIX,gIX,numK,anat_yx,anat_yz,anat_zx,clrmap);
+
+end
+
+function pushbutton_exporttoworkspace_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+assignin('base', 'M', getappdata(hfig,'M'));
+assignin('base', 'cIX', getappdata(hfig,'cIX'));
+assignin('base', 'gIX', getappdata(hfig,'gIX'));
+assignin('base', 'numK', getappdata(hfig,'numK'));
+assignin('base', 'stim', getappdata(hfig,'stim'));
+assignin('base', 'fictive', getappdata(hfig,'fictive'));
+assignin('base', 'M_0_fluo', getappdata(hfig,'M_0_fluo'));
+end
+
+%% row 2: Load
+
+function popup_fishmenu_Callback(hObject,~)
+new_i_fish = get(hObject,'Value');
+hfig = getParentFigure(hObject);
+QuickUpdateFish(hfig,new_i_fish);
+end
+
+function QuickUpdateFish(hfig,new_i_fish,init) %#ok<INUSD>
+CONSTs = getappdata(hfig,'CONSTs');
+setappdata(hfig,'isfullfish',0);
+
+% load all fields from CONSTs, with names preserved
+names = fieldnames(CONSTs{new_i_fish}); % cell of strings
+for i = 1:length(names),
+    setappdata(hfig,names{i},eval(['CONSTs{new_i_fish}.',names{i}]));
+end
+
+% recontruct the 3 big matrices from CIX to original size (numcell)
+CIX = CONSTs{new_i_fish}.CIX;
+numcell = CONSTs{new_i_fish}.numcell;
+
+temp = CONSTs{new_i_fish}.CRAZ;
+CRAZ = zeros(numcell,size(temp,2));
+CRAZ(CIX,:) = temp;
+
+temp = CONSTs{new_i_fish}.CRZt;
+CRZt = zeros(numcell,size(temp,2));
+CRZt(CIX,:) = temp;
+
+temp = CONSTs{new_i_fish}.CInfo;
+CInfo(numcell).center = '';
+CInfo(numcell).area = '';
+CInfo(numcell).slice = '';
+CInfo(numcell).inds = '';
+CInfo(numcell).y_minmax = '';
+CInfo(numcell).x_minmax = '';
+CInfo(CIX) = temp;
+
+setappdata(hfig,'CRAZ',CRAZ); % Matrix array, misc collection
+setappdata(hfig,'CRZt',CRZt); % Matrix array, misc collection
+setappdata(hfig,'CInfo',CInfo); % Matrix array, misc collection
+
+UpdateFishData(hfig,new_i_fish);
+if ~exist('init','var'),
+    UpdateFishDisplay(hfig,new_i_fish);
+end
+end
+
+function UpdateFishData(hfig,new_i_fish)
+UpdateDatamenu_Direct(hfig,1);
+
+% save ClusGroup before updating, if applicable
+clusgroupID = getappdata(hfig,'clusgroupID');
+if ~isempty(clusgroupID),
+    new_clusgroupID = 1;
+    UpdateClusGroupID(hfig,clusgroupID,new_clusgroupID,'norefresh'); % to save ClusGroup
+end
+% set new i_fish after UpdateClusGroupID, which saves into old 'i_fish'
+setappdata(hfig,'i_fish',new_i_fish);
+
+% load from VAR
+global VAR;
+Class = VAR(new_i_fish).Class; % actually collection of classes, each class can have many groups
+setappdata(hfig,'Class',Class);
+
+ClusGroup = VAR(new_i_fish).ClusGroup; % group of 'Cluster', i.e. 2nd level collections
+setappdata(hfig,'ClusGroup',ClusGroup);
+
+clusgroupID = 1;
+setappdata(hfig,'clusgroupID',clusgroupID);
+
+Cluster = ClusGroup{clusgroupID};% collection of clusters, 'Cluster' structurally same as 'Class'
+setappdata(hfig,'Cluster',Cluster);
+end
+
+function UpdateFishDisplay(hfig,new_i_fish)
+stim = getappdata(hfig,'stim');
+M_fish_set = getappdata(hfig,'M_fish_set');
+
+fishset = M_fish_set(new_i_fish);
+[~, names] = GetStimRegressor(stim,fishset);
+global hstimreg;
+set(hstimreg,'String',['(choose)',(names)]);
+
+% update display
+UpdateClassID(hfig,1,'norefresh');
+clusgroupID = 1;
+new_clusgroupID = 1;
+UpdateClusGroupID(hfig,clusgroupID,new_clusgroupID); % to display new menu
+end
+
+function popup_loadfullfishmenu_Callback(hObject,~)
+new_i_fish = get(hObject,'Value')-1;
+if new_i_fish>0,
+    hfig = getParentFigure(hObject);
+    LoadFullFish(hfig,new_i_fish);
+end
+global hfishnum;
+set(hfishnum,'Value',new_i_fish);
+end
+
+function LoadFullFish(hfig,new_i_fish,CONST)
+data_dir=getappdata(hfig,'data_dir');
+if ~exist('CONST','var'),
+    disp(['loading fish #' num2str(new_i_fish) '...']);
+    tic
+    load(fullfile(data_dir,['CONST_F' num2str(new_i_fish) '.mat']),'CONST');
+    toc
+end
+setappdata(hfig,'isfullfish',1);
+
+% load all fields from CONST, with names preserved
+names = fieldnames(CONST); % cell of strings
+for i = 1:length(names),
+    setappdata(hfig,names{i},eval(['CONST.',names{i}]));
+end
+setappdata(hfig,'numcell',length(CONST.CInfo));
+    
+UpdateFishData(hfig,new_i_fish);
+UpdateFishDisplay(hfig,new_i_fish);
+end
+
+function popup_datamenu_Callback(hObject,~)
+ID = get(hObject,'Value');
+hfig = getParentFigure(hObject);
+UpdateDatamenu_Direct(hfig,ID);
+SetDataFR(hfig,1); % also set M (without updating cIX gIX)
+RefreshFigure(hfig);
+end
+
+function UpdateDatamenu_Direct(hfig,ID)
+CRAZ = getappdata(hfig,'CRAZ');
+CRZt = getappdata(hfig,'CRZt');
+FcAvr = getappdata(hfig,'FcAvr');
+Fc = getappdata(hfig,'Fc');
+datanames = getappdata(hfig,'datanames');
+tlists = getappdata(hfig,'tlists');
+photostate = getappdata(hfig,'photostate');
+
+%% SWITCH LEFT/RIGHT DIRECTION AGAIN
+FcAvr = vertcat(FcAvr(2,:),FcAvr(1,:),FcAvr(3:end,:));
+Fc = vertcat(Fc(2,:),Fc(1,:),Fc(3:end,:));
+% now 1=left, 2=right, 3=forward, 4 = raw left, 5 = raw right
+
+%%
+if ID == 1,
+    M_0 = CRAZ;
+    fictive = FcAvr;
+elseif ID == 2,
+    M_0 = CRZt;
+    fictive = Fc;
+else    
+    IX = tlists{ID};
+    M_0 = CRZt(:,IX);
+    temp = Fc(:,IX);
+    % normalize fictive channels, in 2 sets
+    for k = 1:3,%size(F,1),
+        m = temp(k,:);
+        temp(k,:) = (m-min(m))/(max(m)-min(m));
+    end
+    m = temp(4:5,:);
+    temp(4:5,:) = (m-min(min(m)))/(max(max(m))-min(min(m)));
+    fictive = temp;
+end
+% stim from photostate
+stim = photostate(tlists{ID}); 
+
+setappdata(hfig,'M_0_fluo',M_0);
+setappdata(hfig,'fictive',fictive);
+setappdata(hfig,'fictive_0',fictive);
+setappdata(hfig,'dataname',datanames(ID));
+setappdata(hfig,'stim',stim);
+
+global hdatamenu;
+if ~isempty(hdatamenu), % before GUI initialization
+    % 'global' remnant from previous run: hdatamenu is not empty but is invalid object
+    if isvalid(hdatamenu), 
+        set(hdatamenu,'String',datanames);
+        set(hdatamenu,'Value',ID);
+    end
+end
+end
+
+function checkbox_showcentroids_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+setappdata(hfig,'isCentroid',get(hObject,'Value'));
+RefreshFigure(hfig);
+end
+
+%% ----- tab two ----- (Operations)
+
+%% row 1: Range
+
+function pushbutton_back_Callback(hObject,~)
+global hback hfwd;
+hfig = getParentFigure(hObject);
+bC = getappdata(hfig,'bCache');
+fC = getappdata(hfig,'fCache');
+
+if ~isempty(bC.cIX{1}),
+    % set last into forward cache
+    fC.cIX = [getappdata(hfig,'cIX'),fC.cIX];
+    fC.gIX = [getappdata(hfig,'gIX'),fC.gIX];
+    fC.numK = [getappdata(hfig,'numK'),fC.numK];
+    % retrieve
+    cIX = bC.cIX{1};
+    gIX = bC.gIX{1};
+    numK = bC.numK{1};
+    bC.cIX(1) = [];
+    bC.gIX(1) = [];
+    bC.numK(1) = [];
+    % set M
+    M_0 = GetM_0(hfig);
+    M = M_0(cIX,:);
+    % save
+    setappdata(hfig,'M',M);
+    setappdata(hfig,'bCache',bC);
+    setappdata(hfig,'fCache',fC);
+    setappdata(hfig,'cIX',cIX);
+    setappdata(hfig,'gIX',gIX);
+    setappdata(hfig,'numK',numK);
+    % handle rankID: >=2 means write numbers as text next to colorbar
+    setappdata(hfig,'rankID',0);
+    % finish
+    disp('back (from cache)')
+    RefreshFigure(hfig);
+    set(hfwd,'enable','on');
+else % nothing to retrieve
+    set(hback,'enable','off');
+end
+end
+
+function pushbutton_forward_Callback(hObject,~)
+global hback hfwd;
+hfig = getParentFigure(hObject);
+bC = getappdata(hfig,'bCache');
+fC = getappdata(hfig,'fCache');
+
+if ~isempty(fC.cIX{1}),
+    % set last into (backward) cache
+    bC.cIX = [getappdata(hfig,'cIX'),bC.cIX];
+    bC.gIX = [getappdata(hfig,'gIX'),bC.gIX];
+    bC.numK = [getappdata(hfig,'numK'),bC.numK];
+    % retrieve
+    cIX = fC.cIX{1};
+    gIX = fC.gIX{1};
+    numK = fC.numK{1};
+    fC.cIX(1) = [];
+    fC.gIX(1) = [];
+    fC.numK(1) = [];
+    % set M
+    M_0 = GetM_0(hfig);
+    M = M_0(cIX,:);
+    % save
+    setappdata(hfig,'M',M);
+    setappdata(hfig,'bCache',bC);
+    setappdata(hfig,'fCache',fC);
+    setappdata(hfig,'cIX',cIX);
+    setappdata(hfig,'gIX',gIX);
+    setappdata(hfig,'numK',numK);
+    % handle rankID: >=2 means write numbers as text next to colorbar
+    setappdata(hfig,'rankID',0);
+    % finish
+    disp('forward (from cache)')
+    RefreshFigure(hfig);
+    set(hback,'enable','on');
+else
+    set(hfwd,'enable','off');
+end
+end
+
+function edit_choose_range_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+% get/format range
+str = get(hObject,'String'); 
+if ~isempty(str),
+    str = strrep(str,'end',num2str(max(gIX)));
+    range = ParseRange(str);
+    % update indices
+    tempI = [];
+    for i = range,
+        tempI = [tempI;find(gIX==i)];
+    end
+    cIX = cIX(tempI);
+    gIX = gIX(tempI);
+    UpdateIndices(hfig,cIX,gIX);
+    RefreshFigure(hfig);
+end
+end
+
+function range = ParseRange(str)
+str = strrep(str,':','-'); % e.g. str= '1,3,5:8';
+C = textscan(str,'%d','delimiter',',');
+m = C{:};
+range = [];
+for i = 1:length(m),
+    if m(i)>0,
+        range = [range,m(i)]; %#ok<AGROW>
+    else % have '-'sign,
+        range = [range,m(i-1)+1:-m(i)]; %#ok<AGROW>
+    end
+end
+end
+
+function edit_exclude_range_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+% get/format range
+str = get(hObject,'String');
+if ~isempty(str),
+    str = strrep(str,'end',num2str(max(gIX)));
+    range = ParseRange(str);
+    range = setdiff(unique(gIX),range)';
+    % update indices
+    tempI = [];
+    for i = range,
+        tempI = [tempI;find(gIX==i)];
+    end
+    cIX = cIX(tempI);
+    gIX = gIX(tempI);
+    UpdateIndices(hfig,cIX,gIX);
+    RefreshFigure(hfig);
+end
+end
+
+function edit_fuse_range_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+% get/format range
+str = get(hObject,'String');
+if ~isempty(str),
+    str = strrep(str,'end',num2str(max(gIX)));
+    range = ParseRange(str);
+    % update indices
+    tempI = [];
+    for i = range,
+        tempI = [tempI;find(gIX==i)];
+    end
+    gIX(tempI) = gIX(tempI(1));
+    [gIX, numK] = SqueezeGroupIX(gIX);
+    UpdateIndices(hfig,cIX,gIX,numK);
+    RefreshFigure(hfig);
+end
+end
+
+function edit_t_range_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+M = getappdata(hfig,'M');
+
+% get/format range
+str = get(hObject,'String'); 
+if ~isempty(str),
+    str = strrep(str,'end',num2str(size(M,2)));
+    range = ParseRange(str);
+    
+    UpdateTRange(hfig,range)
+    RefreshFigure(hfig);
+end
+end
+
+function UpdateTRange(hfig,range)
+M_0 = GetM_0(hfig);
+cIX = getappdata(hfig,'cIX');
+fictive_0 = getappdata(hfig,'fictive_0');
+% photostate_0 = getappdata(hfig,'photostate_0');
+
+% set M
+M = M_0(cIX,range);
+setappdata(hfig,'M',M);
+% set fictive
+fictive = fictive_0(:,range);
+setappdata(hfig,'fictive',fictive);
+%     % set photostate
+%     photostate = photostate_0(:,range);
+%     setappdata(hfig,'photostate',photostate);
+
+end
+
+%% row 2: Operations
+
+function popup_operations_Callback(hObject,~)
+opID = get(hObject,'Value') - 1;
+hfig = getParentFigure(hObject);
+setappdata(hfig,'opID',opID);
+% highlight UI
+global hopID;
+set(hopID,'BackgroundColor',[1,1,0.8]);
+end
+
+function popup_ranking_Callback(hObject,~)
+% menu = {'(ranking)','hier.','size','stim-lock','corr','noise'};
+rankID = get(hObject,'Value') - 1;
+hfig = getParentFigure(hObject);
+setappdata(hfig,'rankID',rankID);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+M = getappdata(hfig,'M');
+
+[gIX, numU] = SqueezeGroupIX(gIX);
+switch rankID,
+    case 1,
+        disp('hier. (default)');
+        [gIX, numU] = HierClus(M,gIX);
+    case 2,
+        disp('size');
+        H = zeros(numU,1);
+        for i = 1:numU,
+            H(i) = length(find(gIX==i));
+        end
+        [gIX,rankscore] = SortH(H,gIX,numU,'descend');
+    case 3,
+        disp('stim-lock');
+        [gIX,rankscore] = RankByStimLock_Direct(hfig,cIX,gIX,M,numU);
+    case 4,
+        disp('corr');
+        [~,D] = FindCentroid(gIX,M);
+        [gIX,rankscore] = SortH(D,gIX,numU);
+        rankscore = 1-rankscore;
+    case 5,
+        disp('motor');
+        [gIX,rankscore] = RankByMotorStim_Direct(hfig,gIX,M,numU,1);
+    case 6,
+        disp('motor');
+        [gIX,rankscore] = RankByMotorStim_Direct(hfig,gIX,M,numU,2);
+    case 7,
+        disp('motor');
+        [gIX,rankscore] = RankByMotorStim_Direct(hfig,gIX,M,numU,3);
+    case 8,
+        disp('motor');
+        [gIX,rankscore] = RankByMotorStim_Direct(hfig,gIX,M,numU,4);
+    case 9,
+        disp('stim-motor');
+        [~,rankscore1] = RankByStimLock_Direct(hfig,cIX,gIX,M,numU);
+        [~,rankscore2] = RankByMotorStim_Direct(hfig,gIX,M,numU,1);
+        rankscore = rankscore1 - rankscore2;
+end
+if rankID>1,
+    setappdata(hfig,'rankscore',round(rankscore*100)/100);
+    setappdata(hfig,'clrmap','jet');
+else
+    setappdata(hfig,'clrmap','hsv');
+end
+UpdateIndices(hfig,cIX,gIX,numU);
+RefreshFigure(hfig);
+disp('ranking complete');
+end
+
+function [gIX,rankscore] = RankByStimLock_Direct(hfig,cIX,gIX,M,numU)
+periods = getappdata(hfig,'periods');
+
+if length(periods)==1,
+    period = periods{1};
+    [C,~] = FindCentroid(gIX,M);
+else % i_fish>=8,
+    tlists = getappdata(hfig,'tlists');
+    CRZt = getappdata(hfig,'CRZt');
+    IX = tlists{6}; % ptomr_circ
+    M_ = CRZt(cIX,IX);
+    [C,~] = FindCentroid(gIX,M_);
+    periods = getappdata(hfig,'periods');
+    period = periods{1}+periods{2};
+end
+% C2 = zscore(C,0,2);
+% C_3D = reshape(C2,size(C2,1),period,[]);
+C_3D_0 = reshape(C,size(C,1),period,[]);
+C_3D = zscore(C_3D_0,0,2);
+
+H = nanmean(nanstd(C_3D,0,3),2);
+[gIX,rankscore] = SortH(H,gIX,numU);
+end
+
+function [gIX,rankscore] = RankByMotorStim_Direct(hfig,gIX,M,numU,option)
+C = FindCentroid(gIX,M);
+fictive = getappdata(hfig,'fictive');
+
+reg = zeros(3,length(fictive));
+reg(1,:) = fictive(5,:); % Left
+reg(2,:) = fictive(4,:); % Right
+reg(3,:) = mean(fictive(4:5,:));
+H = zeros(numU,1);
+shift = zeros(numU,1);
+a = zeros(1,3);
+I = zeros(1,3);
+for i = 1:numU,
+    switch option,
+        case 1,
+            for j = 1:3,
+                [a(j),I(j)] = max(abs(xcorr(C(i,:),reg(j,:),'coeff')));
+            end
+            [H(i),jmax] = max(a);
+            shift(i) = I(jmax) - length(fictive);
+        case 2,
+            [H(i),I] = max(xcorr(C(i,:),reg(1,:),'coeff'));
+            shift(i) = I - length(fictive);
+        case 3,
+            [H(i),I] = max(xcorr(C(i,:),reg(2,:),'coeff'));
+            shift(i) = I - length(fictive);
+        case 4,
+            [H(i),I] = max(xcorr(C(i,:),reg(3,:),'coeff'));
+            shift(i) = I - length(fictive);
+    end
+end
+[gIX,rankscore] = SortH(H,gIX,numU,'descend');
+assignin('base', 'shift', shift);
+end
+
+function [gIX,B] = SortH(H,gIX,numU,descend) % new gIX is sorted based on H, size(H)=[numU,1];
+if exist('descend','var'),
+    [B,I] = sort(H,'descend');
+else
+    [B,I] = sort(H);
+end
+gIX_last = gIX;
+for i = 1:numU,
+    gIX(gIX_last==I(i)) = i;
+end
+end
+
+function pushbutton_sqeeze_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+[gIX, numU] = SqueezeGroupIX(gIX);
+UpdateIndices(hfig,cIX,gIX,numU);
+RefreshFigure(hfig);
+end
+
+function pushbutton_flipud_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+
+U = unique(gIX);
+U_ = flipud(U);
+
+gIX_ = gIX;
+for i = 1:length(U),
+    gIX_(gIX==U(i)) = U_(i);
+end
+
+UpdateIndices(hfig,cIX,gIX_);
+RefreshFigure(hfig);
+end
+
+function pushbutton_clrmap_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+clrmap = getappdata(hfig,'clrmap');
+if strcmp(clrmap,'jet'),
+    setappdata(hfig,'clrmap','hsv');
+else
+    setappdata(hfig,'clrmap','jet');
+end
+RefreshFigure(hfig);
+end
+
+%% row 3: Anatomy
+
+function pushbutton_polygon_yx_Callback(hObject,~)
+k_zres = 20;
+
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+% M = getappdata(hfig,'M');
+CInfo = getappdata(hfig,'CInfo');
+numcell = getappdata(hfig,'numcell');
+anat_yx = getappdata(hfig,'anat_yx');
+anat_zx = getappdata(hfig,'anat_zx');
+dimv_yx = size(anat_yx);
+dimv_zx = size(anat_zx);
+isfullfish = getappdata(hfig,'isfullfish');
+
+h_poly_yx = impoly;
+wait(h_poly_yx); % double click to finalize position!
+% update finalized polygon in bright color
+setColor(h_poly_yx,[0 1 1]);
+
+if isfullfish,
+    IJs = reshape([CInfo.center],2,[])';
+else % Matlab can't handle CInfo with all the empty entries -> manual padding
+    IJs_ = reshape([CInfo(cIX).center],2,[])';
+    IJs = ones(numcell,2);
+    IJs(cIX,:) = IJs_;
+end
+A = sub2ind(dimv_yx(1:2),IJs(:,1),IJs(:,2));
+MaskArray = createMask(h_poly_yx);
+MaskArray(1:dimv_zx*k_zres+10,:) = [];
+B = find(MaskArray); % find indices of pixels within ROI
+
+temp = find(ismember(A,B));
+[IX,ia,~] = intersect(cIX,temp);
+
+cIX = IX;
+gIX = gIX(ia);
+
+UpdateIndices(hfig,cIX,gIX);
+RefreshFigure(hfig);
+end
+
+function pushbutton_polygon_yz_Callback(hObject,~)
+k_zres = 20;
+
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+% M = getappdata(hfig,'M');
+CInfo = getappdata(hfig,'CInfo');
+numcell = getappdata(hfig,'numcell');
+anat_yx = getappdata(hfig,'anat_yx');
+anat_yz = getappdata(hfig,'anat_yz');
+anat_zx = getappdata(hfig,'anat_zx');
+dimv_yx = size(anat_yx);
+dimv_yz = size(anat_yz);
+dimv_zx = size(anat_zx);
+isfullfish = getappdata(hfig,'isfullfish');
+
+h_poly_z = impoly;
+wait(h_poly_z); % double click to finalize position!
+% update finalized polygon in bright color
+setColor(h_poly_z,[0 1 1]);
+
+if isfullfish,
+        IJs = reshape([CInfo.center],2,[])';
+        Zs = [CInfo.slice]';
+else % Matlab can't handle CInfo with all the empty entries -> manual padding
+    IJs_ = reshape([CInfo(cIX).center],2,[])';
+    IJs = ones(numcell,2);
+    IJs(cIX,:) = IJs_;
+    Zs_ = [CInfo(cIX).slice]';
+    Zs = ones(numcell,1);
+    Zs(cIX) = Zs_;
+end
+A = sub2ind(dimv_yz(1:2),IJs(:,1),Zs);
+
+MaskArray = createMask(h_poly_z);
+MaskArray(1:dimv_zx*k_zres+10,:) = [];
+MaskArray(:,1:dimv_yx(2)+10) = [];
+zMaskArray = imresize(MaskArray,[dimv_yz(1),dimv_yz(2)]);
+B = find(zMaskArray); % find indices of pixels within ROI
+
+temp = find(ismember(A,B));
+[IX,ia,~] = intersect(cIX,temp);
+
+cIX = IX;
+gIX = gIX(ia);
+
+UpdateIndices(hfig,cIX,gIX);
+RefreshFigure(hfig);
+end
+
+function pushbutton_polygon_zx_Callback(hObject,~)
+k_zres = 20;
+
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+% M = getappdata(hfig,'M');
+CInfo = getappdata(hfig,'CInfo');
+numcell = getappdata(hfig,'numcell');
+% anat_yx = getappdata(hfig,'anat_yx');
+anat_zx = getappdata(hfig,'anat_zx');
+% dimv_yx = size(anat_yx);
+dimv_zx = size(anat_zx);
+isfullfish = getappdata(hfig,'isfullfish');
+
+h_poly_z = impoly;
+wait(h_poly_z); % double click to finalize position!
+% update finalized polygon in bright color
+setColor(h_poly_z,[0 1 1]);
+
+if isfullfish,
+        IJs = reshape([CInfo.center],2,[])';
+        Zs = [CInfo.slice]';
+else % Matlab can't handle CInfo with all the empty entries -> manual padding
+    IJs_ = reshape([CInfo(cIX).center],2,[])';
+    IJs = ones(numcell,2);
+    IJs(cIX,:) = IJs_;
+    Zs_ = [CInfo(cIX).slice]';
+    Zs = ones(numcell,1);
+    Zs(cIX) = Zs_;
+end
+A = sub2ind(dimv_zx(1:2),Zs,IJs(:,2));
+
+MaskArray = createMask(h_poly_z);
+
+MaskArray(dimv_zx(1)*k_zres+1:end,:) = [];
+MaskArray(:,dimv_zx(2)+1:end) = [];
+zMaskArray = imresize(MaskArray,[dimv_zx(1),dimv_zx(2)]);
+zMaskArray = flipud(zMaskArray); % careful!
+B = find(zMaskArray); % find indices of pixels within ROI
+
+temp = find(ismember(A,B));
+[IX,ia,~] = intersect(cIX,temp);
+
+cIX = IX;
+gIX = gIX(ia);
+
+UpdateIndices(hfig,cIX,gIX);
+RefreshFigure(hfig);
+end
+
+function pushbutton_withinConvexHull_Callback(hObject,~) % only works with full load
+disp('find points within boundary...');
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+CInfo = getappdata(hfig,'CInfo');
+IJs = reshape([CInfo.center],2,[])';
+Zs = [CInfo.slice]';
+% point sets
+xyz_all = horzcat(IJs, Zs);
+xyz_cIX = horzcat(IJs(cIX,:), Zs(cIX));
+
+tri = delaunayn(xyz_cIX); % Generate delaunay triangulization
+t = tsearchn(xyz_cIX, tri, xyz_all); % Determine which triangle point is within
+I_inside = ~isnan(t);
+
+cIX_1 = cIX;
+gIX_1 = gIX;
+cIX_2 = setdiff(find(I_inside),cIX_1);
+gIX_2 = ones(length(cIX_2),1)*(max(gIX_1)+1);
+cIX = [cIX_1;cIX_2];
+gIX = [gIX_1;gIX_2];
+
+UpdateIndices(hfig,cIX,gIX,length(unique(gIX)));
+RefreshFigure(hfig);
+disp('search complete');
+end
+
+%% ----- tab three ----- (Regression)
+
+%% row 1: regressor
+
+function popup_getstimreg_Callback(hObject,~)
+i_reg = get(hObject,'Value')-1;
+hfig = getParentFigure(hObject);
+if i_reg==0,
+    return;
+end
+setappdata(hfig,'regchoice',{1,i_reg});
+% highlight the choice (yellow)
+global hstimreg hmotorreg hcentroidreg;
+set(hstimreg,'BackgroundColor',[1,1,0.8]); % yellow
+set(hmotorreg,'BackgroundColor',[1,1,1]);
+set(hcentroidreg,'BackgroundColor',[1,1,1]);
+
+isPlotReg = getappdata(hfig,'isPlotReg');
+if isPlotReg,
+    PlotRegWithStimMotor(hfig);
+end
+end
+
+function PlotRegWithStimMotor(hfig)
+stim = getappdata(hfig,'stim');
+fictive = getappdata(hfig,'fictive');
+regressor = GetRegressor(hfig);
+% plot regressor
+figure;
+halfbarheight = 1;
+stimbar = GetStimBar(halfbarheight,stim);
+reg = imNormalize99(regressor);
+regbar = repmat(reg,[1,1,3]);
+
+subplot(3,1,1);image(stimbar); axis off; title('stimulus');
+subplot(3,1,2);image(regbar); axis off; title('regressor');
+subplot(3,1,3);imagesc(fictive);colormap hot; axis off; title('motor');
+end
+
+function out=imNormalize99(im)
+im=double(im);
+temp=sort(im(:),'descend');
+th1=temp(round(length(im(:))/100));
+th2=min(im(:));
+
+out=(im-th2)/(th1-th2);
+out(out>1)=1;
+end
+
+function popup_getmotorreg_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+i_reg = get(hObject,'Value')-1;
+% {'left swims','right swims','forward swims','raw left','raw right','raw average'};
+if i_reg>0,
+    setappdata(hfig,'regchoice',{2,i_reg});
+end
+% highlight the choice
+global hstimreg hmotorreg hcentroidreg;
+set(hstimreg,'BackgroundColor',[1,1,1]);
+set(hmotorreg,'BackgroundColor',[1,1,0.8]); % yellow
+set(hcentroidreg,'BackgroundColor',[1,1,1]);
+
+isPlotReg = getappdata(hfig,'isPlotReg');
+if isPlotReg,
+    PlotRegWithStimMotor(hfig);
+end
+end
+
+function checkbox_popupplotreg_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+setappdata(hfig,'isPlotReg',get(hObject,'Value'));
+end
+
+function edit_ctrdID_as_reg_Callback(hObject,~)
+str = get(hObject,'String');
+if ~isempty(str),
+    temp = textscan(str,'%d');
+    ctrdID = temp{:};
+end
+hfig = getParentFigure(hObject);
+setappdata(hfig,'regchoice',{3,ctrdID});
+% highlight the choice
+global hstimreg hmotorreg hcentroidreg;
+set(hstimreg,'BackgroundColor',[1,1,1]);
+set(hmotorreg,'BackgroundColor',[1,1,1]);
+set(hcentroidreg,'BackgroundColor',[1,1,0.8]); % yellow
+end
+
+function checkbox_flipstim_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+setappdata(hfig,'isflipstim',get(hObject,'Value'));
+end
+
+%% row 2: regression
+
+function regressor = GetRegressor(hObject)
+hfig = getParentFigure(hObject);
+regchoice = getappdata(hfig,'regchoice');
+stim = getappdata(hfig,'stim');
+    
+if regchoice{1}==1, % stim Regressor
+    i_fish = getappdata(hfig,'i_fish');
+    M_fish_set = getappdata(hfig,'M_fish_set');
+    fishset = M_fish_set(i_fish);
+    [regressors, names] = GetStimRegressor(stim,fishset);    
+    regressor = regressors(regchoice{2}).im;
+    
+elseif regchoice{1}==2, % motor Regressor
+    fictive = getappdata(hfig,'fictive');
+    regressors = GetMotorRegressor(fictive);
+    regressor = regressors(regchoice{2}).im;
+    
+else % regchoice{1}==3, from Centroid
+    ctrdID = regchoice{2};
+    isflipstim = getappdata(hfig,'isflipstim');
+    M = getappdata(hfig,'M');
+    gIX = getappdata(hfig,'gIX');
+    
+    i = find(unique(gIX)==ctrdID);
+    if isempty(i),
+        disp('input is empty!');beep;
+        regressor = [];
+        return;
+    end
+    [C,~] = FindCentroid(gIX,M);
+    regressor = C(i,:);
+    
+    if isflipstim, % swap left/right stim
+        % 0 = all black; 1 = black/white; 2 = white/black; 3 = all white; 4 = all gray;
+        % 10 = forward grating (very slow, more for calibration)
+        % 11 = rightward grating
+        % 12 = leftward grating
+        % swap: 1~2,11~12
+        
+        %     if i_fish == 6 || i_fish == 7,
+        %         halflength = length(stim)/2;
+        %         stim = stim(1:halflength); % length is always even
+        %     end
+        flipstim = stim;
+        flipstim(stim==1)=2;
+        flipstim(stim==2)=1;
+        flipstim(stim==11)=12;
+        flipstim(stim==12)=11;
+        
+        phototrans = GetPhotoTrans(stim);
+        fliptrans = GetPhotoTrans(flipstim);
+        
+        % transition e.g.:
+        % photostate: 2     3     1     3     3     0     0     1     1     0     3     2     0     2     2     1
+        % phototrans: 6    11    13     7    15    12     0     1     5     4     3    14     8     2    10     9
+        
+        IX = [];
+        targetstates = unique(fliptrans,'stable');
+        for i = 1:length(targetstates),
+            IX = [IX, find(phototrans==targetstates(i))];
+        end
+        %     if i_fish == 6 || i_fish == 7,
+        %         reg = regressor(1:halflength);
+        %         reg = reg(IX);
+        %         regressor = repmat(reg,1,2);
+        %     else
+        regressor = regressor(IX);
+        %     end
+    end    
+end
+
+end
+
+function checkbox_plotcorrhist_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+setappdata(hfig,'isPlotCorrHist',get(hObject,'Value'));
+end
+
+function pushbutton_topnum_regression_Callback(hObject,~)
+disp('regression...');
+hfig = getParentFigure(hObject);
+M_0 = getappdata(hfig,'M_0_fluo');
+numTopCorr = getappdata(hfig,'numTopCorr');
+regressor = GetRegressor(hfig);
+
+%% for each cell, find correlation coeff
+M_corr = corr(regressor',M_0');%cell_resp_ave');
+% M_corr_ctrl = corr(regressor.ctrl',M_0');
+
+[~,I] = sort(M_corr,'descend');
+cIX = I(1:numTopCorr)';
+gIX = ceil((1:length(cIX))'/length(cIX)*min(20,length(cIX)));
+
+score = M_corr(cIX);
+rankscore = zeros(1,20);
+for i = 1:20,
+    ix = find(gIX==i);
+    rankscore(i) = mean(score(ix));
+end
+setappdata(hfig,'rankscore',round(rankscore*100)/100);
+setappdata(hfig,'rankID',99);
+
+UpdateIndices(hfig,cIX,gIX,20);
+RefreshFigure(hfig);
+end
+
+function edit_topCorrNumber_Callback(hObject,~)
+str = get(hObject,'String');
+if ~isempty(str),
+    temp = textscan(str,'%d');
+    numTopCorr = temp{:};
+    hfig = getParentFigure(hObject);
+    setappdata(hfig,'numTopCorr',numTopCorr);
+end
+end
+
+function pushbutton_thres_regression_Callback(hObject,~) 
+disp('regression...');
+hfig = getParentFigure(hObject);
+M_0 = getappdata(hfig,'M_0_fluo');
+thres_reg = getappdata(hfig,'thres_reg');
+regressor = GetRegressor(hfig);
+if isempty(regressor),
+    return;
+end
+
+isPlotCorrHist = getappdata(hfig,'isPlotCorrHist');
+
+[cIX,gIX] = Regression_direct(M_0,thres_reg,regressor,isPlotCorrHist);
+
+if ~isempty(gIX),
+    gIX = ceil((1:length(cIX))'/length(cIX)*min(20,length(cIX)));
+    UpdateIndices(hfig,cIX,gIX,40);
+    RefreshFigure(hfig);
+end
+end
+
+function [cIX_,gIX_,R] = Regression_direct(M_0,thres_reg,regressor,isPlotCorrHist) % gIX_ is just ones
+R = corr(regressor',M_0');
+if thres_reg>0,
+    cIX_ = (find(R>thres_reg))';
+elseif thres_reg<0,
+    cIX_ = (find(R<thres_reg))';
+end
+if isempty(cIX_),
+    disp('result is empty!');beep;
+    gIX_ = [];
+    return;
+end
+[~,I] = sort(R(cIX_),'descend');
+cIX_ = cIX_(I);
+gIX_ = ones(size(cIX_));
+%%
+if exist('isPlotCorrHist','var'),
+    if isPlotCorrHist,
+        figure('Position',[500,200,500,150]);
+        hold on;
+        bins = -1:0.025:1;
+        [N,~] = histcounts(R,bins);
+        histogram(R,bins,'FaceColor',[0.4 0.4 0.4]);%,'EdgeColor','none'
+        plot([thres_reg,thres_reg],[0,max(N)],'r--');
+        xlim([-1,1]);ylim([0,max(N)]);
+    end
+end
+end
+
+function edit_regthres_Callback(hObject,~)
+str = get(hObject,'String');
+if ~isempty(str),
+    temp = textscan(str,'%f');
+    thres_reg = temp{:};
+    hfig = getParentFigure(hObject);
+    setappdata(hfig,'thres_reg',thres_reg);
+end
+end
+
+function pushbutton_AllCentroidRegression_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+M_0 = getappdata(hfig,'M_0_fluo');
+thres_reg = getappdata(hfig,'thres_reg');
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+
+[cIX,gIX,numK] = AllCentroidRegression_Direct(M_0,thres_reg,cIX,gIX);
+
+UpdateIndices(hfig,cIX,gIX,numK);
+RefreshFigure(hfig);
+
+disp('all regression complete');
+end
+
+function [cIX,gIX,numK] = AllCentroidRegression_Direct(M_0,thres_reg,cIX,gIX)
+U = unique(gIX);
+clusters(length(U)).cIX = [];
+for i = 1:length(U),
+    disp(['i = ' num2str(i)]);
+    IX = find(gIX == U(i));    
+    [~,regressor] = kmeans(M_0(cIX(IX),:),1,'distance','correlation'); %#ok<FNDSB>
+    cIX_ = Regression_direct(M_0,thres_reg,regressor);
+    clusters(i).cIX = cIX_;
+end
+disp('union...');
+CIX = vertcat(clusters(1:end).cIX);
+GIX = [];
+for i = 1:numel(clusters),
+    GIX = [GIX; ones(length(clusters(i).cIX),1)*i]; %#ok<AGROW>
+end
+[cIX,gIX,numK] = SmartUnique(CIX,GIX,M_0(CIX,:));
+end
+
+function pushbutton_IterCentroidRegression_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+M_0 = getappdata(hfig,'M_0_fluo');
+thres_reg = getappdata(hfig,'thres_reg');
+M = getappdata(hfig,'M');
+gIX = getappdata(hfig,'gIX');
+regchoice = getappdata(hfig,'regchoice');
+if regchoice{1}==3, % from Centroid
+    ctrdID = regchoice{2};
+end
+
+i = find(unique(gIX)==ctrdID);
+if isempty(i),
+    disp('input is empty!');beep;
+    return;
+end
+[C,~] = FindCentroid(gIX,M);
+regressor = C(i,:);
+[cIX_,gIX_] = Regression_direct(M_0,thres_reg,regressor);
+numC = length(cIX_);
+regressor_last = regressor;
+
+if ~isempty(cIX_),
+    for itr = 1:20,
+        [cIX_,gIX_] = Regression_direct(M_0,thres_reg,regressor_last);
+        numC = length(cIX_);
+        disp(num2str(numC));
+        M = M_0(cIX_,:);
+        regressor = FindCentroid(gIX_,M);
+        if itr>1 && abs(numC-numC_last)<5,%corr(regressor_last',regressor')>thres_iter,
+            disp(['converged at itr = ' num2str(itr)]);
+            break;
+        end
+        regressor_last = regressor;
+        numC_last = numC;
+    end
+    gIX_ = ceil((1:length(cIX_))'/length(cIX_)*min(20,length(cIX_)));
+    UpdateIndices(hfig,cIX_,gIX_,40);
+    RefreshFigure(hfig);
+end
+end
+
+%% row 3: (TBD)
+
+function checkbox_dataFR_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+SetDataFR(hfig,get(hObject,'Value'));
+RefreshFigure(hfig);
+end
+
+%% ----- tab four ----- (Clustering etc.)
+
+%% row 1: k-means
+
+function edit_kmeans_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+M = getappdata(hfig,'M');
+
+str = get(hObject,'String');
+if ~isempty(str),
+    temp = textscan(str,'%d');
+    numK = temp{:};
+    disp(['k-means k=' num2str(numK) '...']);
+    tic
+    rng('default');% default = 0, but can try different seeds if doesn't converge
+    if numel(M)*numK < 10^7 && numK~=1,
+        disp('Replicates = 5');
+        [gIX,C,sumd,D] = kmeans(M,numK,'distance','correlation','Replicates',5);
+    elseif numel(M)*numK < 10^8 && numK~=1,
+        disp('Replicates = 3');
+        [gIX,C,sumd,D] = kmeans(M,numK,'distance','correlation','Replicates',3);
+    else
+        [gIX,C,sumd,D] = kmeans(M,numK,'distance','correlation');%,'Replicates',3);
+    end
+    toc
+    beep
+    
+    if numK>1,
+        gIX = HierClusDirect(C,gIX,numK);
+    end
+    
+    cIX = getappdata(hfig,'cIX');
+    UpdateIndices(hfig,cIX,gIX,numK);
+    RefreshFigure(hfig);
+end
+end
+
+function edit_kmeans2_Callback(hObject,~) % based on anatomical distance
+hfig = getParentFigure(hObject);
+z_res = getappdata(hfig,'z_res');
+cIX = getappdata(hfig,'cIX');
+CInfo = getappdata(hfig,'CInfo');
+M = getappdata(hfig,'M');
+
+XY = reshape([CInfo(cIX).center],2,[])';
+Z = [CInfo(cIX).slice]'.*z_res;
+XYZ = horzcat(XY,Z);
+Combo = horzcat(XYZ/50,M);% how to weight????????????????
+
+str = get(hObject,'String');
+if ~isempty(str),
+    temp = textscan(str,'%d');
+    numK = temp{:};
+    disp(['anat. weighted k-means k=' num2str(numK) '...']);
+    tic
+    rng('default');% default = 0, but can try different seeds if doesn't converge
+    [gIX,C,sumd,D] = kmeans(Combo,numK,'distance','correlation');%,'Replicates',3);
+    toc
+    beep
+    
+    if numK>1,
+        gIX = HierClusDirect(C,gIX,numK);
+    end
+
+    cIX = getappdata(hfig,'cIX');
+    UpdateIndices(hfig,cIX,gIX,numK);
+    RefreshFigure(hfig);
+end
+end
+
+function edit_kmeans_elbow_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+gIX = getappdata(hfig,'gIX');
+M = getappdata(hfig,'M');
+
+str = get(hObject,'String');
+if ~isempty(str),
+    str = strrep(str,'end',num2str(max(gIX)));
+    range = ParseRange(str);
+
+    Silh_mean = zeros(1,length(range));
+    for i = 1:length(range),
+        disp(['k-means k=' num2str(range(i))]);
+        gIX = kmeans(M,range(i),'distance','correlation');        
+        silh = silhouette(M,gIX,'correlation');
+        Silh_mean(i) = mean(silh);
+    end
+    [~,ix] = max(Silh_mean);    
+    numK = range(ix);
+    
+    figure;hold on
+    plot(range,Silh_mean,'o-','color',[0.5 0.5 0.5]);
+    
+    if length(range)~=1,
+        plot([numK,numK],[min(Silh_mean),max(Silh_mean)],'r--');
+        xlim([range(1),range(end)]);
+        ylim([min(Silh_mean),max(Silh_mean)])
+    end
+    xlabel('k = # of clusters')
+    ylabel('silhouette mean')
+    
+    %% perform regular k-means
+    disp(['k-means k=' num2str(numK)]);
+    rng('default');% default = 0, but can try different seeds if doesn't converge
+    if numel(M)*numK < 10^7 && numK~=1,
+        disp('Replicates = 5');
+        [gIX,C,sumd,D] = kmeans(M,numK,'distance','correlation','Replicates',5);
+    elseif numel(M)*numK < 10^8 && numK~=1,
+        disp('Replicates = 3');
+        [gIX,C,sumd,D] = kmeans(M,numK,'distance','correlation','Replicates',3);
+    else
+        [gIX,C,sumd,D] = kmeans(M,numK,'distance','correlation');%,'Replicates',3);
+    end       
+    if numK>1,
+        gIX = HierClusDirect(C,gIX,numK);
+    end    
+    cIX = getappdata(hfig,'cIX');    
+    UpdateIndices(hfig,cIX,gIX,numK);
+    RefreshFigure(hfig);
+    beep;
+end
+end
+
+%% row 2: Auto-clustering
+
+function pushbutton_merge_Callback(hObject,~)
+% disp('merging...');
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+M = getappdata(hfig,'M');
+U = unique(gIX);
+numU = length(U);
+[C,D] = FindCentroid(gIX,M);
+
+thres_merge = getappdata(hfig,'thres_merge');
+
+i = 1;
+while i<numU,    
+    c = corr(C(i,:)',C(i+1,:)');
+    if c > thres_merge,
+        IX = find(gIX == U(i+1));
+        gIX(IX)=U(i);
+        U = unique(gIX);
+        numU = length(U);
+        
+        IX = find(gIX == U(i));
+        M_s = M(IX,:);
+        [~,C1,~,D1] = kmeans(M_s,1,'distance','correlation');
+        C(i,:) = C1;
+        D(i) = mean(D1);
+        C(i+1,:) = [];
+        D(i+1) = [];
+    else
+        i = i+1;
+    end
+end
+
+if numU>1,
+    [gIX, numU] = HierClus(M,gIX);
+end
+
+UpdateIndices(hfig,cIX,gIX,numU);
+RefreshFigure(hfig);
+disp('merging complete');
+end
+
+function edit_mergethres_Callback(hObject,~)
+str = get(hObject,'String');
+temp = textscan(str,'%f',1);
+hfig = getParentFigure(hObject);
+setappdata(hfig,'thres_merge',temp{:});
+end
+
+function pushbutton_iter_split(hObject,~)
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+numU = getappdata(hfig,'numK');
+thres_split = getappdata(hfig,'thres_split');
+M_0 = getappdata(hfig,'M_0_fluo');
+classheader = getappdata(hfig,'classheader');
+
+disp('iter. split all, beep when done...');
+thres_size = 10;
+thres_H = thres_split; 
+% thres_H = [0.2;0.15;0.1;0.05]; % could have more rounds...
+
+% initialization
+I_rest = [];
+% loop
+tic
+for round = 1:length(thres_H),
+    disp(['round ' num2str(round) ', numU = ' num2str(numU)]);
+    dthres = 1-thres_H(round);
+    gIX_last = gIX;
+    I_clean_last = cIX;
+    cIX = [];
+    gIX = [];
+    for i = 1:numU,
+        disp(['i = ' num2str(i)]);
+        ix = find(gIX_last == i);
+%         IX = ix;
+        IX = I_clean_last(ix);
+        M_s = M_0(IX,:);
+        [I_rest,cIX,gIX,numU] = CleanClus(M_s,IX,I_rest,cIX,gIX,numU,dthres,thres_size);
+%         cIX = I_clean_last(I_clean);
+    end
+
+    [gIX, numU] = SqueezeGroupIX(gIX);
+    SaveClass(hfig,cIX,gIX,classheader,['clean_round' num2str(round)]);
+
+    SaveClass(hfig,I_rest,ones(length(I_rest),1),classheader,...
+       ['rest_round' num2str(round)]);
+end
+toc
+beep
+end
+
+function [I_rest,I_clean,gIX_clean,numU] = CleanClus(M_s,IX,I_rest,I_clean,gIX_clean,numU,dthres,thres_size)
+I_clean_s = [];
+
+% find numK_s for kmeans
+kmax = min(round(size(M_s,1)/thres_size),30);
+% try numK_s = 1
+numK_s = 1;
+rng('default');
+[gIX_s,~,~,D] = kmeans(M_s,numK_s,'distance','correlation');
+Dist = min(D,[],2);
+if mean(Dist)>dthres,
+    % try numK_s = kmax
+    numK_s = kmax;
+    rng('default');
+    [gIX_s,~,~,D] = kmeans(M_s,numK_s,'distance','correlation');
+    Dist = min(D,[],2);
+    if mean(Dist)<dthres, % find in between value for numK_s
+        numK_s = 2;
+        while 1,
+            % kmeans-cluster by numK_s
+            rng('default');
+            [gIX_s,~,~,D] = kmeans(M_s,numK_s,'distance','correlation');
+            Dist = min(D,[],2);
+            if mean(Dist)<dthres,
+                break;
+            end
+            
+            if numK_s < kmax,
+                numK_s = numK_s+1;
+                disp(['numK_s = ' num2str(numK_s)]);
+            else % numK_s = kmax;
+                break;
+            end
+        end
+    else disp(['numK_s = ' num2str(numK_s)]);
+    end
+end
+% have numK_s that makes mean(Dist) < thres, or numK_s = kmax
+
+for i = 1:numK_s,
+    IX_s = find(gIX_s == i);
+    if length(IX_s)>thres_size, % cluster big enough to start
+        dst = Dist(IX_s);
+        if mean(dst) < dthres,
+            I_clean_s = [I_clean_s; IX_s];
+            gIX_clean = [gIX_clean; gIX_s(IX_s)+double(numU)];
+        else
+            ix = find(dst<dthres); % clean
+            if length(ix)>=thres_size, % clean cluster still big enough
+                I = IX_s(ix);
+                I_clean_s = [I_clean_s; I];
+                gIX_clean = [gIX_clean; gIX_s(I)+double(numU)];
+            end
+        end
+    end
+end
+
+numU = numU + numK_s;
+I_rest_s = setdiff(1:size(M_s,1),I_clean_s);
+I_rest = [I_rest; IX(I_rest_s)];
+I_clean = [I_clean; IX(I_clean_s)];
+end
+
+function edit_splitthres_Callback(hObject,~)
+str = get(hObject,'String');
+temp = textscan(str,'%f',1);
+hfig = getParentFigure(hObject);
+setappdata(hfig,'thres_split',temp{:});
+end
+
+function pushbutton_autoclus_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+M = getappdata(hfig,'M');
+M_0 = getappdata(hfig,'M_0_fluo');
+classheader = getappdata(hfig,'classheader');
+thres_size = 10;
+thres_split = getappdata(hfig,'thres_split');
+thres_stimlock = 1.0;
+thres_merge = getappdata(hfig,'thres_merge');
+thres_silh = 0.4;
+
+isWkmeans = getappdata(hfig,'isWkmeans');
+
+%% kmeans
+if isWkmeans,
+    numK = 20;
+    disp(['kmeans k = ' num2str(numK)]);
+    tic
+    rng('default');% default = 0, but can try different seeds if doesn't converge
+    if numel(M)*numK < 10^7 && numK~=1,
+        disp('Replicates = 5');
+        gIX = kmeans(M,numK,'distance','correlation','Replicates',5);
+    elseif numel(M)*numK < 10^8 && numK~=1,
+        disp('Replicates = 3');
+        gIX = kmeans(M,numK,'distance','correlation','Replicates',3);
+    else
+        gIX = kmeans(M,numK,'distance','correlation');
+    end
+    toc
+end
+[gIX, numU] = SqueezeGroupIX(gIX);
+
+%% pushbutton_iter_split(hObject,~);
+disp('iter. split all...');
+I_rest = [];
+iter = 1;
+gIX_last = gIX;
+I_clean_last = cIX;
+cIX = [];
+gIX = [];
+for i = 1:numU,
+    disp(['i = ' num2str(i)]);
+    ix = gIX_last == i;
+    IX = I_clean_last(ix);
+    M_s = M_0(IX,:);
+    [I_rest,cIX,gIX,numU] = CleanClus(M_s,IX,I_rest,cIX,gIX,numU,1-thres_split,thres_size);
+end
+[gIX, ~] = SqueezeGroupIX(gIX);
+if isempty(gIX),
+    errordlg('nothing to display!');
+    return;
+end
+SaveClass(hfig,cIX,gIX,classheader,['clean_round' num2str(iter)]);
+SaveClass(hfig,I_rest,ones(length(I_rest),1),classheader,...
+    ['rest_round' num2str(iter)]);
+
+[gIX, numU] = Merge_direct(thres_merge,M_0,cIX,gIX);
+
+%% rank by stim-lock
+disp('stim-lock');
+M = M_0(cIX,:);
+[gIX,rankscore] = RankByStimLock_Direct(hfig,cIX,gIX,M,numU);
+disp('ranking complete');
+% and threshold
+IX = find(rankscore<thres_stimlock);
+ix = ismember(gIX,IX);
+gIX = gIX(ix);
+cIX = cIX(ix);
+
+%% ~ pushbutton_autoregclus_Callback(hObject,~);
+thres_reg = getappdata(hfig,'thres_reg');
+[cIX,gIX,~] = AllCentroidRegression_Direct(M_0,thres_reg,cIX,gIX);
+disp('auto-reg-clus complete');
+
+[gIX, numU] = Merge_direct(thres_merge,M_0,cIX,gIX);
+
+SaveClass(hfig,cIX,gIX,classheader,'clean_round2');
+
+%% Silhouette
+disp('silhouette analysis');
+gIX_last = gIX;
+for i = 1:numU,
+    disp(['i = ' num2str(i)]);
+    IX = find(gIX_last == i);
+    cIX_2 = cIX(IX);
+    M_s = M_0(cIX_2,:);
+    % try k-means with k=2, see whether to keep
+    gIX_ = kmeans(M_s,2,'distance','correlation');
+    silh = silhouette(M_s,gIX_,'correlation');
+    if mean(silh)>thres_silh,
+        % keep the k-means k=2 subsplit
+        disp('split');
+        gIX(IX) = gIX_ + numU; % reassign (much larger) gIX
+    end
+end
+[gIX, ~] = SqueezeGroupIX(gIX);
+
+SaveClass(hfig,cIX,gIX,classheader,'clean_round3');
+
+%% rank by stim-lock ?? bug?
+% disp('stim-lock');
+% M = M_0(cIX,:);
+% [gIX,rankscore] = RankByStimLock_Direct(hfig,cIX,gIX,M,numU);
+% disp('ranking complete');
+% % and threshold
+% IX = find(rankscore<thres_stimlock);
+% ix = ismember(gIX,IX);
+% gIX = gIX(ix);
+% cIX = cIX(ix);
+% 
+% [gIX, ~] = Merge_direct(thres_merge,M_0,cIX,gIX);
+
+% size threshold
+thres_size = getappdata(hfig,'thres_size');
+[cIX, gIX, numU] = ThresSize(cIX,gIX,thres_size);
+
+%% update GUI
+if isempty(gIX),
+    errordlg('nothing to display!');
+    return;
+end
+UpdateIndices(hfig,cIX,gIX,numU);
+RefreshFigure(hfig);
+beep;
+
+end
+
+function pushbutton_thressize_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+thres_size = getappdata(hfig,'thres_size');
+[cIX, gIX, numU] = ThresSize(cIX,gIX,thres_size);
+UpdateIndices(hfig,cIX,gIX,numU);
+RefreshFigure(hfig);
+end
+
+function edit_sizethres_Callback(hObject,~)
+str = get(hObject,'String');
+if ~isempty(str),
+    temp = textscan(str,'%f');
+    thres_size = temp{:};
+end
+hfig = getParentFigure(hObject);
+setappdata(hfig,'thres_size',thres_size);
+end
+
+function [cIX, gIX, numU] = ThresSize(cIX,gIX,thres_size)
+U = unique(gIX);
+numU = length(U);
+for i=1:numU,
+    if length(find(gIX==U(i)))<thres_size,
+        cIX(gIX==U(i)) = [];
+        gIX(gIX==U(i)) = [];
+    end
+end
+[gIX, numU] = SqueezeGroupIX(gIX);
+end
+
+function [gIX, numU] = Merge_direct(thres_merge,M_0,cIX,gIX)
+M = M_0(cIX,:);
+[gIX, numU] = HierClus(M,gIX);
+U = unique(gIX);
+M = M_0(cIX,:);
+[C,D] = FindCentroid(gIX,M);
+i = 1;
+while i<numU,    
+    c = corr(C(i,:)',C(i+1,:)');
+    if c > thres_merge,
+        IX = find(gIX == U(i+1));
+        gIX(IX)=U(i);
+        U = unique(gIX);
+        numU = length(U);
+        
+        IX = find(gIX == U(i));
+        M_s = M(IX,:);
+        [~,C1,~,D1] = kmeans(M_s,1,'distance','correlation');
+        C(i,:) = C1;
+        D(i) = mean(D1);
+        C(i+1,:) = [];
+        D(i+1) = [];
+    else
+        i = i+1;
+    end
+end
+[gIX, numU] = HierClus(M,gIX);
+disp('merging complete');
+end
+
+function checkbox_wkmeans_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+setappdata(hfig,'isWkmeans',get(hObject,'Value'));
+end
+
+%% row 3: misc plots
+
+function pushbutton_hierplot_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+M = getappdata(hfig,'M');
+
+[gIX2, numU] = HierClus(M,gIX,'isplotfig');
+if ~isequal(gIX,gIX2),
+    UpdateIndices(hfig,cIX,gIX2,numU);
+    RefreshFigure(hfig);
+end
+end
+
+function edit_hierpartn_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+M = getappdata(hfig,'M');
+gIX = getappdata(hfig,'gIX');
+hierinplace = getappdata(hfig,'hierinplace');
+
+[gIX, numU] = SqueezeGroupIX(gIX);
+[C,~] = FindCentroid(gIX,M);
+
+str = get(hObject,'String');
+if ~isempty(str),
+    temp = textscan(str,'%d',1);
+    numCuts = temp{:};
+    
+%     tree = linkage(C,'average','correlation');
+%     IX_tree = cluster(tree,'maxclust',numCuts);
+    IX_tree = clusterdata(C,'criterion','distance','distance','correlation','maxclust',numCuts)
+    
+    if hierinplace,
+        % sort to keep clusters in place
+        U = unique(IX_tree,'stable');
+        temp = zeros(size(IX_tree));
+        for i = 1:length(U),
+            temp(IX_tree==U(i)) = i;
+        end
+        IX_tree = temp;
+    end
+    
+    % update gIX
+    temp = zeros(size(gIX));
+    for i = 1:numU,
+        temp(gIX==i) = IX_tree(i);
+    end
+    gIX = temp;
+
+    cIX = getappdata(hfig,'cIX');
+    UpdateIndices(hfig,cIX,gIX,length(unique(IX_tree)));
+    RefreshFigure(hfig);
+end
+end
+
+function edit_hierpartthres_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+M = getappdata(hfig,'M');
+gIX = getappdata(hfig,'gIX');
+hierinplace = getappdata(hfig,'hierinplace');
+
+[gIX, numU] = SqueezeGroupIX(gIX);
+[C,~] = FindCentroid(gIX,M);
+
+str = get(hObject,'String');
+if ~isempty(str),
+    temp = textscan(str,'%f',1);
+    thres= temp{:};
+    
+    IX_tree = clusterdata(C,'criterion','distance',...
+        'distance','correlation','cutoff',thres);
+        
+    if hierinplace,
+        % sort to keep clusters in place
+        U = unique(IX_tree,'stable');
+        temp = zeros(size(IX_tree));
+        for i = 1:length(U),
+            temp(IX_tree==U(i)) = i;
+        end
+        IX_tree = temp;
+    end
+    
+    % update gIX
+    temp = zeros(size(gIX));
+    for i = 1:numU,
+        temp(gIX==i) = IX_tree(i);
+    end
+    gIX = temp;
+
+    cIX = getappdata(hfig,'cIX');
+    UpdateIndices(hfig,cIX,gIX,length(unique(IX_tree)));
+    RefreshFigure(hfig);
+end
+end
+
+function checkbox_hierinplace_Callback(hObject,~)
+hierinplace = get(hObject,'Value');
+hfig = getParentFigure(hObject);
+setappdata(hfig,'hierinplace',hierinplace);
+end
+
+function pushbutton_corrplot_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+gIX = getappdata(hfig,'gIX');
+M = getappdata(hfig,'M');
+[C,~] = FindCentroid(gIX,M);
+coeffs = corr(C');%corr(C(1,:)',C(2,:)')
+CorrPlot(coeffs);
+end
+
+function CorrPlot(coeffs)
+im = coeffs;
+
+% red-white-blue colormap
+cmap = zeros(64,3);
+cmap(:,1) = [linspace(0,1,32), linspace(1,1,32)];
+cmap(:,2) = [linspace(0,1,32), linspace(1,0,32)];
+cmap(:,3) = [linspace(1,1,32), linspace(1,0,32)];
+minlim = -1; %min(min(im));
+maxlim = 1; %max(max(im));
+
+RGB = ImageToRGB(im,cmap,minlim,maxlim); % map image matrix to range of colormap
+
+figure('Position',[1000,200,500,500]);
+image(RGB); axis equal; axis tight;
+
+for i = 1:size(im,2), % horizontal.. because of image axis
+    for j = 1:size(im,1),
+        text(i-0.3, j, num2str(round(im(i,j)*100)/100));%, 'Units', 'data')
+    end
+end
+end
+
+function RGB = ImageToRGB(im,cmap,minlim,maxlim)
+L = size(cmap,1);
+ix = round(interp1(linspace(minlim,maxlim,L),1:L,im,'linear','extrap'));
+RGB = reshape(cmap(ix,:),[size(ix) 3]); % Make RGB image from scaled.
+end
+
+function pushbutton_plot4x4_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+PlotBy16StimsFromGUI(hfig);
+end
+
+%% ----- tab five ----- (Saved Clusters)
+
+%% row 1: Class
+
+function popup_classmenu_Callback(hObject,~)
+classID = get(hObject,'Value') - 1;
+if classID>0,
+    hfig = getParentFigure(hObject);
+    UpdateClassID(hfig,classID); % update popupmenu
+end
+end
+
+function edit_editclassname_Callback(hObject,~)
+str = get(hObject,'String');
+
+hfig = getParentFigure(hObject);
+Class = getappdata(hfig,'Class');
+classID = getappdata(hfig,'classID');
+Class(classID).name = str;
+setappdata(hfig,'Class',Class);
+UpdateClassID(hfig,classID); % update popupmenu
+end
+
+function pushbutton_saveclass_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+SaveClass(hfig);
+end
+
+function edit_class_header_Callback(hObject,~)
+% get/format range
+str = get(hObject,'String');
+
+hfig = getParentFigure(hObject);
+setappdata(hfig,'classheader',[str ':']);
+end
+
+function edit_newclassname_Callback(hObject,~)
+str = get(hObject,'String');
+hfig = getParentFigure(hObject);
+setappdata(hfig,'newclassname',str);
+end
+
+function pushbutton_makeclass_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+classheader = getappdata(hfig,'classheader');
+newclassname = getappdata(hfig,'newclassname');
+SaveClass(hfig,cIX,gIX,classheader,newclassname);
+end
+
+function pushbutton_delclass_Callback(hObject,~)
+choice = questdlg('Delete current class?','','Cancel','Yes','Yes');
+if strcmp(choice,'Yes'), 
+    hfig = getParentFigure(hObject);
+    Class = getappdata(hfig,'Class');
+    classID = getappdata(hfig,'classID');
+    disp(['delete ' num2str(classID)]);
+    Class(classID) = [];
+    setappdata(hfig,'Class',Class);
+        
+    classID = max(1,classID-1);
+    disp(['new classID: ' num2str(classID)]);
+    UpdateClassID(hfig,classID);
+end
+end
+
+%% row 2: Cluster
+
+function popup_clusgroupmenu_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+clusgroupID = getappdata(hfig,'clusgroupID');
+new_clusgroupID = get(hObject,'Value');
+UpdateClusGroupID(hfig,clusgroupID,new_clusgroupID);
+end
+
+function popup_clusmenu_Callback(hObject,~)
+clusID = get(hObject,'Value') - 1;
+if clusID>0,
+    hfig = getParentFigure(hObject);
+    UpdateClusID(hfig,clusID);
+end
+end
+
+function edit_editclusname_Callback(hObject,~)
+str = get(hObject,'String');
+
+hfig = getParentFigure(hObject);
+Cluster = getappdata(hfig,'Cluster');
+clusID = getappdata(hfig,'clusID');
+Cluster(clusID).name = str;
+setappdata(hfig,'Cluster',Cluster);
+UpdateClusID(hfig,clusID); % update popupmenu
+end
+
+function pushbutton_saveclus_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+SaveCluster(hfig,'current');
+end
+
+function edit_clus_header_Callback(hObject,~)
+% get/format range
+str = get(hObject,'String');
+
+hfig = getParentFigure(hObject);
+setappdata(hfig,'clusheader',[str ':']);
+end
+
+function edit_newclusname_Callback(hObject,~)
+str = get(hObject,'String');
+hfig = getParentFigure(hObject);
+setappdata(hfig,'newclusname',str);
+end
+
+function pushbutton_makeclus_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+SaveCluster(hfig,'new');
+end
+
+function edit_setrank_Callback(hObject,~)
+str = get(hObject,'String');
+C = textscan(str,'%d');
+rank = C{:};
+
+hfig = getParentFigure(hObject);
+Cluster = getappdata(hfig,'Cluster');
+clusID = getappdata(hfig,'clusID');
+% insert current cluster into new position = 'rank'
+if rank > 1,
+    temp = Cluster(clusID);
+    Cluster(clusID) = [];
+    Cluster = [Cluster(1:rank-1),temp,Cluster(rank:end)];
+else 
+    temp = Cluster(clusID);
+    Cluster(clusID) = [];
+    Cluster = [temp,Cluster(rank:end)];
+end
+setappdata(hfig,'Cluster',Cluster);
+% update pointer = clusID
+clusID = rank;
+UpdateClusID(hfig,clusID);
+end
+
+function edit_notes_Callback(hObject,~)
+str = get(hObject,'String');
+hfig = getParentFigure(hObject);
+Cluster = getappdata(hfig,'Cluster');
+clusID = getappdata(hfig,'clusID');
+Cluster(clusID).notes = str;
+setappdata(hfig,'Cluster',Cluster);
+end
+
+function pushbutton_delclus_Callback(hObject,~)
+choice = questdlg('Delete current cluster?','','Cancel','Yes','Yes');
+if strcmp(choice,'Yes'), 
+    hfig = getParentFigure(hObject);
+    Cluster = getappdata(hfig,'Cluster');
+    clusID = getappdata(hfig,'clusID');
+    disp(['delete ' num2str(clusID)]);
+    Cluster(clusID) = [];
+    setappdata(hfig,'Cluster',Cluster);
+        
+    clusID = max(1,clusID-1);
+    disp(['new clusID: ' num2str(clusID)]);
+    UpdateClusID(hfig,clusID);
+end
+end
+
+%% row 3: misc
+
+function edit_clusUnion_Callback(hObject,~)
+disp('union processing...');
+hfig = getParentFigure(hObject);
+M_0 = GetM_0(hfig);
+Cluster = getappdata(hfig,'Cluster');
+str = get(hObject,'String');
+if ~isempty(str),
+    % get/format range
+    str = strrep(str,':','-'); % e.g. str= '1,3,5:8';
+    C = textscan(str,'%d','delimiter',',');
+    m = C{:};
+    range = [];
+    for i = 1:length(m),
+        if m(i)>0,
+            range = [range,m(i)]; %#ok<AGROW>
+        else % have '-'sign,
+            range = [range,m(i-1)+1:-m(i)]; %#ok<AGROW>
+        end
+    end
+
+    % combine
+    CIX = vertcat(Cluster(range).cIX);
+    GIX = []; % gIX to match A
+    for i = 1:length(range),
+        GIX = [GIX; ones(length(Cluster(range(i)).gIX),1)*i];
+    end    
+    [cIX,gIX,numK] = SmartUnique(CIX,GIX,M_0(CIX,:));    
+    UpdateIndices(hfig,cIX,gIX,numK);    
+    RefreshFigure(hfig);
+end
+disp('union complete');
+end
+
+function [cIX,gIX,numK] = SmartUnique(CIX,GIX,M) % input: simply concatenated groups
+disp('unique based on corr coeff...');
+CTRD = FindCentroid(GIX,M); 
+
+[uCIX,ia,~] = unique(CIX);
+uGIX = GIX(ia);
+if length(uCIX) == 1, 
+    counts = length(x);
+else counts = hist(CIX,uCIX); 
+end
+IX1 = find(counts==1);
+CIX1 = uCIX(IX1); % single occurence
+GIX1 = uGIX(IX1);
+ia = find(~ismember(CIX,CIX1));
+C = CIX(ia); % multiple copies/occurence, keep all copies
+G = GIX(ia);
+M_s = M(ia,:);
+Coeff = zeros(length(ia),1);
+for i = 1:length(ia),
+    reg = CTRD(G(i),:);
+    Coeff(i) = corr(M_s(i,:)',reg');
+end
+[~,I] = sort(Coeff,'descend'); % rank by coeff, so unique ('stable') gets highest
+C = C(I); % finish sorting
+G = G(I);
+[CIX2,ia,~] = unique(C,'stable'); % keep the order!
+GIX2 = G(ia); % the chosen copy for those with multiple copies
+cIX = vertcat(CIX1,CIX2);
+gIX = vertcat(GIX1,GIX2);
+[gIX, numK] = SqueezeGroupIX(gIX);
+disp('smart union complete');
+end
+
+function pushbutton_newclusgroup_Callback(hObject,~)
+hfig = getParentFigure(hObject);
+ClusGroup = getappdata(hfig,'ClusGroup');
+ClusGroup = [ClusGroup,cell(1,1)];
+setappdata(hfig,'ClusGroup',ClusGroup);
+
+clusgroupID = getappdata(hfig,'clusgroupID');
+new_clusgroupID = length(ClusGroup);
+UpdateClusGroupID(hfig,clusgroupID,new_clusgroupID);
+end
+
+function pushbutton_delclusgroup_Callback(hObject,~)
+choice = questdlg('Delete current Cluster-group?','','Cancel','Yes','Yes');
+if strcmp(choice,'Yes'),
+    hfig = getParentFigure(hObject);
+    ClusGroup = getappdata(hfig,'ClusGroup');
+    clusgroupID = getappdata(hfig,'clusgroupID');
+    ClusGroup(clusgroupID) = [];
+    setappdata(hfig,'ClusGroup',ClusGroup);
+    UpdateClusGroupID(hfig,[],max(1,clusgroupID-1));
+end
+end
+
+%% Internal functions
+
+function SetDataFR(hfig,dataFR)
+cIX = getappdata(hfig,'cIX');
+setappdata(hfig,'dataFR',dataFR);
+if dataFR,
+    M_0_fluo = getappdata(hfig,'M_0_fluo');
+    setappdata(hfig,'M',M_0_fluo(cIX,:));
+else % Unchecked
+    M_0_reg = getappdata(hfig,'M_0_reg');
+    setappdata(hfig,'M',M_0_reg(cIX,:));
+end
+global hdataFR;
+set(hdataFR,'Value',dataFR);
+end
+
+function SaveClass(hfig,cIX,gIX,classheader,newclassname)
+if ~exist('cIX','var'),
+    cIX = getappdata(hfig,'cIX');
+end
+if ~exist('gIX','var'),
+    gIX = getappdata(hfig,'gIX');
+end
+
+Class = getappdata(hfig,'Class');
+dataname = getappdata(hfig,'dataname');
+
+if ~exist('newclassname','var'),
+    classID = getappdata(hfig,'classID');
+else
+    classID = numel(Class)+1;
+    Class(classID).name = [classheader newclassname];
+end
+
+Class(classID).cIX = cIX;
+Class(classID).gIX = gIX;
+Class(classID).numel = length(cIX);
+U = unique(gIX);
+numU = length(U);
+Class(classID).numK = numU;
+Class(classID).dataname = dataname;
+
+setappdata(hfig,'Class',Class);
+UpdateClassID(hfig,classID); 
+disp('class saved');
+end
+
+function UpdateClassID(hfig,classID,norefresh)
+Class = getappdata(hfig,'Class');
+numK = Class(classID).numK;
+i_fish = getappdata(hfig,'i_fish');
+% save
+setappdata(hfig,'classID',classID);
+% update GUI
+global hclassname hclassmenu;
+if ishandle(hclassname),
+    set(hclassname,'String',Class(classID).name);
+    temp = MakeMenu({Class.name});
+    set(hclassmenu, 'String', temp);
+    set(hclassmenu, 'Value', classID+1);
+end
+
+if ~exist('norefresh','var'),
+    UpdateIndices(hfig,Class(classID).cIX,Class(classID).gIX,numK);
+    RefreshFigure(hfig);
+end
+global VAR;
+VAR(i_fish).Class = getappdata(hfig,'Class');
+end
+
+function UpdateClusGroupID(hfig,clusgroupID,new_clusgroupID,norefresh)
+% save/update old Cluster into ClusGroup before exiting, 
+% as Cluster is the variable handled in hfig but not saved elsewhere
+ClusGroup = getappdata(hfig,'ClusGroup');
+Cluster = getappdata(hfig,'Cluster');
+i_fish = getappdata(hfig,'i_fish');
+
+ClusGroup{clusgroupID} = Cluster;
+setappdata(hfig,'ClusGroup',ClusGroup);
+global VAR;
+VAR(i_fish).ClusGroup = CurrentClusGroup(hfig);
+
+Cluster = ClusGroup{new_clusgroupID};
+setappdata(hfig,'Cluster',Cluster);
+setappdata(hfig,'clusgroupID',new_clusgroupID);
+
+global hclusgroupmenu;
+if ishandle(hclusgroupmenu),
+    num = length(ClusGroup);
+    menu = {}; for i = 1:num, menu = [menu,{num2str(i)}];end
+    set(hclusgroupmenu,'String', menu,'Value',new_clusgroupID);
+end
+
+if ~exist('norefresh','var'),    
+    if numel(Cluster) == 0,
+        SaveCluster(hfig,'new');
+    else
+        clusID = 1;
+        UpdateClusID(hfig,clusID);
+    end
+end
+end
+
+function ClusGroup = CurrentClusGroup(hfig)
+ClusGroup = getappdata(hfig,'ClusGroup');
+Cluster = getappdata(hfig,'Cluster');
+clusgroupID = getappdata(hfig,'clusgroupID');
+ClusGroup{clusgroupID} = Cluster;
+setappdata(hfig,'ClusGroup',ClusGroup);
+end
+
+function [Cluster,clusID] = SaveCluster(hfig,state,clusheader,name)
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+
+Cluster = getappdata(hfig,'Cluster');
+dataname = getappdata(hfig,'dataname');
+
+if strcmp(state,'current'),
+    clusID = getappdata(hfig,'clusID');
+else %if strcmp(state,'new'),
+    clusID = numel(Cluster)+1;
+    if ~exist('name','var'),
+        name = getappdata(hfig,'newclusname');
+    end
+    if ~exist('clusheader','var'),
+        clusheader = getappdata(hfig,'clusheader');
+    end
+    Cluster(clusID).name = [clusheader name];
+end
+
+Cluster(clusID).cIX = cIX;
+Cluster(clusID).gIX = gIX;
+Cluster(clusID).numel = length(cIX);
+U = unique(gIX);
+numU = length(U);
+Cluster(clusID).numK = numU;
+Cluster(clusID).dataname = dataname;
+
+setappdata(hfig,'Cluster',Cluster);
+UpdateClusID(hfig,clusID); 
+disp('cluster saved');
+end
+
+function UpdateClusID(hfig,clusID)
+Cluster = getappdata(hfig,'Cluster');
+    
+% save
+setappdata(hfig,'clusID',clusID);
+% update GUI
+global hclusname hclusmenu;
+set(hclusname,'String',Cluster(clusID).name);
+menu = MakeMenu({Cluster.name});
+set(hclusmenu,'String', menu,'Value',clusID+1);
+numK = Cluster(clusID).numK;
+
+UpdateIndices(hfig,Cluster(clusID).cIX,Cluster(clusID).gIX,numK);
+RefreshFigure(hfig);
+end
+
+function menu = MakeMenu(name) % careful: input {Class.name}, otherwise only get 1 somehow
+menu = [{'(choose)'},name];
+for j=2:length(menu),menu(j)={[num2str(j-1) ': ' menu{j}]};end
+end
+
+function [gIX, numU] = HierClus(M,gIX,isplotfig) %#ok<INUSD>
+[gIX, numU] = SqueezeGroupIX(gIX);
+[C,~] = FindCentroid(gIX,M);
+D = pdist(C,'correlation');
+tree = linkage(C,'average','correlation');
+leafOrder = optimalleaforder(tree,D);
+
+if numU>1,
+    if exist('isplotfig','var'),
+            figure('Position',[100 400 1300 400]);
+            subplot(1,3,1);
+            CORR = corr(C');
+            imagesc(CORR);axis equal;axis tight
+            
+            subplot(1,3,2);
+            dendrogram(tree,numU,'orientation','right','reorder',leafOrder);
+            set(gca,'YDir','reverse');
+            colormap(jet);
+
+            subplot(1,3,3);
+            C2 = C(leafOrder,:);
+            CORR2 = corr(C2');
+            imagesc(CORR2);axis equal;axis tight
+    end
+    % sort for uniform colorscale
+    temp = zeros(size(gIX));
+    for i = 1:numU,
+        temp(gIX==leafOrder(i)) = i; % = T(i) for clusters segmented from tree
+    end
+    gIX = temp;
+end
+end
+
+function gIX = HierClusDirect(C,gIX,numU)
+D = pdist(C,'correlation');
+tree = linkage(C,'average','correlation');
+leafOrder = optimalleaforder(tree,D);
+
+% sort for uniform colorscale
+temp = zeros(size(gIX));
+for i = 1:numU,
+    temp(gIX==leafOrder(i)) = i; % = T(i) for clusters segmented from tree
+end
+gIX = temp;
+end
+
+function [gIX, numK] = SqueezeGroupIX(gIX)
+U = unique(gIX);
+numK = length(U);
+for i = 1:numK,
+    old = U(i);
+    gIX(gIX==old) = i;
+end
+end
+
+% frequently used, updates cell-index,group-index,cluster-number. set-operations included in here.
+function UpdateIndices(hfig,cIX,gIX,numK)
+global hback hopID;
+M_0 = GetM_0(hfig);
+if ~exist('gIX','var'),
+    gIX = getappdata(hfig,'gIX');
+end
+if ~exist('cIX','var'),
+    cIX = getappdata(hfig,'cIX');
+end
+
+% update cache
+bC = getappdata(hfig,'bCache');
+cIX_last = getappdata(hfig,'cIX');
+gIX_last = getappdata(hfig,'gIX');
+if ~(isequal(cIX_last,cIX) && isequal(gIX_last,gIX)),
+    bC.cIX = [cIX_last,bC.cIX];
+    bC.gIX = [gIX_last,bC.gIX];
+    bC.numK = [getappdata(hfig,'numK'),bC.numK];
+    set(hback,'enable','on');
+    if length(bC.cIX)>20,
+        bC.cIX(end) = [];
+        bC.gIX(end) = [];
+        bC.numK(end) = [];
+    end
+end
+
+% set operations, if applicable
+opID = getappdata(hfig,'opID');
+if opID ~= 0,
+    switch opID,
+        case 1,
+            disp('union');
+            [~,ia,ib] = union(cIX_last,cIX,'stable');
+            IX = vertcat(cIX_last(ia),cIX(ib));% IX;
+        case 2,
+            disp('intersect');
+            [IX,ia,~] = intersect(cIX_last,cIX);
+            ib = [];
+        case 3,
+            disp('setdiff');
+            [IX,ia] = setdiff(cIX_last,cIX);
+            ib = [];
+        case 4,
+            disp('rev setdiff');
+            % swap sequence, then same as opID==3
+            temp = cIX;
+            cIX = cIX_last;
+            cIX_last = temp;
+            temp = gIX;
+            gIX = gIX_last;
+            gIX_last = temp;
+            [IX,ia] = setdiff(cIX_last,cIX);
+            ib = [];
+        case 5,
+            disp('setxor');
+            [IX,ia,ib] = setxor(cIX_last,cIX);
+        case 6,
+            disp('smartUnion');
+            CIX = vertcat(cIX_last,cIX);
+            GIX = [gIX_last;gIX+max(gIX_last)]; % gIX to match
+            [cIX,gIX,numK] = SmartUnique(CIX,GIX,M_0(CIX,:));              
+    end
+    if opID<6,
+        if ~isempty(IX),
+            cIX = IX;
+            gIX = vertcat(gIX_last(ia),gIX(ib)+max(gIX_last(ia)));
+            numK = length(unique(gIX));
+            %         [gIX, numK] = SqueezeGroupIX(gIX);
+        else
+            errordlg('operation result is empty set!')
+            waitforbuttonpress;
+        end
+    end
+    set(hopID,'Value',1,'BackgroundColor',[1,1,1]); % reset
+    setappdata(hfig,'opID',0);
+end
+
+if length(cIX)*size(M_0,2)<5*10^8,
+    % set M
+    M = M_0(cIX,:);
+    
+    setappdata(hfig,'M',M);
+    setappdata(hfig,'bCache',bC);
+    setappdata(hfig,'cIX',cIX);
+    setappdata(hfig,'gIX',gIX);
+    if exist('numK','var'),
+        setappdata(hfig,'numK',numK);
+    end
+else
+    errordlg('dataset too large!')
+    waitforbuttonpress;
+end
+
+% handle rankID: >=2 means write numbers as text next to colorbar
+% first UpdateIndices sets rankID to 100, second sets back to 0 
+rankID = getappdata(hfig,'rankID');
+if rankID>=2,
+    if rankID==100,
+        setappdata(hfig,'rankID',0);
+    else
+        setappdata(hfig,'rankID',100);
+    end
+end
+end
+
+% frequently used, 2 plotting functions are outside ('DrawClusters.m' and 'DrawClustersOnMap_LSh.m')
+function RefreshFigure(hfig)
+CInfo = getappdata(hfig,'CInfo');
+M = getappdata(hfig,'M');
+dataFR = getappdata(hfig,'dataFR');
+fictive = getappdata(hfig,'fictive');
+cIX = getappdata(hfig,'cIX');
+gIX = getappdata(hfig,'gIX');
+
+numK = getappdata(hfig,'numK');
+stim = getappdata(hfig,'stim');
+anat_yx = getappdata(hfig,'anat_yx');
+anat_yz = getappdata(hfig,'anat_yz');
+anat_zx = getappdata(hfig,'anat_zx');
+isCentroid = getappdata(hfig,'isCentroid');
+clrmap = getappdata(hfig,'clrmap');
+rankscore = getappdata(hfig,'rankscore');
+rankID = getappdata(hfig,'rankID');
+
+iswrite = (rankID>=2);
+
+if isempty(cIX),
+    errordlg('empty set! go back!');
+    return;
+end
+
+allAxesInFigure = findall(hfig,'type','axes');
+if ~isempty(allAxesInFigure)
+    delete(allAxesInFigure);
+end
+
+figure(hfig);
+% left subplot
+h1 = axes('Position',[0.05, 0.04, 0.55, 0.83]);
+if isCentroid,
+    [C,~] = FindCentroid(gIX,M);
+    DrawClusters(h1,C,unique(gIX),dataFR,numK,stim,fictive,clrmap,rankscore,iswrite);
+else
+    DrawClusters(h1,M,gIX,dataFR,numK,stim,fictive,clrmap,rankscore,iswrite);
+end
+
+% right subplot
+h2 = axes('Position',[0.61, 0.04, 0.35, 0.83]);
+DrawClustersOnMap_LSh(CInfo,cIX,gIX,numK,anat_yx,anat_yz,anat_zx,clrmap);
+end
+
+function M_0 = GetM_0(hfig)
+dataFR = getappdata(hfig,'dataFR');
+if dataFR,
+    M_0 = getappdata(hfig,'M_0_fluo');
+else
+    M_0 = getappdata(hfig,'M_0_reg');
+end
+end
+
+function [C,D] = FindCentroid(gIX,M)
+U = unique(gIX);
+numU = length(U);
+C = zeros(numU,size(M,2));
+D = zeros(numU,1);
+for i=1:numU,
+    IX = find(gIX == U(i));
+    if length(IX)==1,
+        C(i,:) = M(IX,:);
+        D(i) = 1;
+    else
+        M_s = M(IX,:);
+        [~,C1,~,D1] = kmeans(M_s,1,'distance','correlation');
+        C(i,:) = C1;
+        D(i) = mean(D1);
+    end
+end
+end
+
+function closefigure_Callback(hfig,~)
+global EXPORT_autorecover;
+EXPORT_autorecover = getappdata(hfig);
+end
+
+function fig = getParentFigure(fig)
+% if the object is a figure or figure descendent, return the figure. Otherwise return [].
+while ~isempty(fig) && ~strcmp('figure', get(fig,'type'))
+  fig = get(fig,'parent');
+end
+end
+
+function runscript(flag_script,var_script)
+switch flag_script
+    case 'push_cIX_gIX'
+        UpdateIndices(var_script{:});
+        RefreshFigure(var_script{1});        
+end
+end
+% end
diff --git a/first version/GetMotorRegressor.m b/first version/GetMotorRegressor.m
new file mode 100644
index 0000000..73f17e7
--- /dev/null
+++ b/first version/GetMotorRegressor.m	
@@ -0,0 +1,86 @@
+function regressors = GetMotorRegressor(fictive)
+%% generate GCaMP6f kernel
+% GCaMP6f: decay half-time: 400±41; peak: 80±35
+% GCaMP6s: 1796±73, 480±24
+
+fpsec = 1.97;
+
+tlen=size(fictive,2);
+t=0:0.05:8; % sec
+gc6=exp(-(t-(4+0.48))/1.8);
+gc6(t<(4+0.48))=0;
+t_im = 0:1/fpsec:1/fpsec*(tlen-1);
+t_gc6=0:0.05:tlen; % sec
+%%
+%     turn(data.swimStartIndT(i)-100,1)  = turn_directLeft(i);     %binary: left turns
+%     turn(data.swimStartIndT(i)-100,2)  = turn_directRight(i);    %binary: right turns
+%     turn(data.swimStartIndT(i)-100,3)  = turn_direct(i);         %binary: all turns (+1 for left -1 for right)
+%     turn(data.swimStartIndT(i)-100,4)  = forward_direct(i);      %binary: forward swims
+%     turn(data.swimStartIndT(i)-100,5)  = swim_direct(i);         %binary: all swims
+%     
+%     turn(data.swimStartIndT(i)-100,6)  = turn_amp(i);            %weighted: all turns
+%     turn(data.swimStartIndT(i)-100,7)  = turn_left(i);           %weighted: left turns
+%     turn(data.swimStartIndT(i)-100,8)  = turn_right(i);          %weighted: right turns
+%     turn(data.swimStartIndT(i)-100,9)  = forward_amp(i);         %weighted: forward swims
+%     turn(data.swimStartIndT(i)-100,10) = swim_amp(i);            %weighted: all swims
+%     
+%     turn(data.swimStartIndT(i)-100,11) = left_amp(i);            %weighted: left channel % 11,12 similar to 10?!
+%     turn(data.swimStartIndT(i)-100,12) = right_amp(i);           %weighted: right channel 
+%     turn(:,13)  = fltCh1;                                            %analog: left channel   
+%     turn(:,14)  = fltCh2;                                            %analog: right channel   
+regressor_0={ % rows = [7,8,9,13,14];
+    fictive(1,:);   %weighted: right turns
+    fictive(2,:);   %weighted: left turns
+    fictive(3,:);   %weighted: forward swims
+    fictive(4,:);  %analog: right channel
+    fictive(5,:);  %analog: left channel   
+    fictive(4,:)+fictive(5,:);   %analog: average
+    };
+nRegType = length(regressor_0);
+name_array = {'w_right','w_left','w_fwd','raw_right','raw_left','raw_all'};
+
+% segment length and round off, for shuffled control
+segLength = floor(tlen/80);
+s4_ = tlen-mod(tlen,segLength);
+
+% initialize/preallocate struct
+regressors(nRegType).name = [];
+regressors(nRegType).im = [];
+regressors(nRegType).ctrl = [];
+
+%% feed all regressors into struct
+idx = 0;
+for j=1:nRegType, %run_StimRegType_subset,
+    len = size(regressor_0{j},1);
+    for i=1:len, %run_PhotoState_subset,
+        idx = idx + 1;
+        reg = regressor_0{j}(i,:);
+        %         idx = (j-1)*nStimRegType + i;
+        
+        regressors(idx).name = [name_array{j} '_' num2str(i)];
+        % generate regressor in imaging time
+        regressors(idx).im = gen_reg_im(gc6, t_gc6, t_im, reg);
+        
+        % generate shuffled regressor
+        reg_ = reg(1:s4_);
+        reg2D = reshape(reg_, segLength, []);
+        indices = randperm(size(reg2D,2)); % reshuffle by indexing
+        shffreg = reg2D(:, indices);
+        shffreg = reshape(shffreg,1,[]);
+        temp = reg;
+        temp(1:s4_) = shffreg;
+        shffreg = temp;
+        regressors(idx).ctrl = gen_reg_im(gc6, t_gc6, t_im, shffreg);
+        
+    end
+end
+
+end
+
+function reg_im = gen_reg_im(gc6, t_gc6, t_im, reg)
+
+temp1=interp1(t_im,reg,t_gc6,'linear','extrap');
+temp2=conv(temp1,gc6,'same');
+reg_im = interp1(t_gc6,temp2,t_im,'linear','extrap');
+
+end
diff --git a/GetPhotoTrans.m b/first version/GetPhotoTrans.m
similarity index 100%
rename from GetPhotoTrans.m
rename to first version/GetPhotoTrans.m
diff --git a/first version/GetStimBar.m b/first version/GetStimBar.m
new file mode 100644
index 0000000..2f7215d
--- /dev/null
+++ b/first version/GetStimBar.m	
@@ -0,0 +1,25 @@
+function stimbar = GetStimBar(halfbarheight,stim) % horizontal stimulus bar. nFrames is length of actual content, barlength is raw length including padding
+m1 = 0.3*ones(halfbarheight,length(stim));
+m2 = m1; % bottom half
+x = stim;
+
+% 0 = all black; 1 = black/white; 2 = white/black; 3 = all white; 4 = all gray;
+% 5 = gray/black; 6 = white/gray; 7 = black/gray; 8 = gray/white.
+% 10 = forward grating (very slow, more for calibration)
+% 11 = rightward grating
+% 12 = leftward grating
+
+% top (~ projection left)
+m1(:,x==0 | x==2 | x==5) = 0; % black
+m1(:,x==4 | x==6 | x==7) = 0.8; % grey
+m1(:,x==1 | x==3 | x==8) = 1; % white
+% bottom (~ projection right)
+m2(:,x==0 | x==1 | x==7) = 0;
+m2(:,x==4 | x==5 | x==8) = 0.8;
+m2(:,x==2 | x==3 | x==6) = 1;
+% colors for OMR
+stimbar = repmat(vertcat(m1,m2),[1 1 3]); % 3 color layers in dim3
+stimbar(:,x==11,1) = 1; % red ~ left
+stimbar(:,x==10,2) = 1; % green ~ forward
+stimbar(:,x==12,3) = 1; % blue ~ right
+end
diff --git a/first version/GetStimRegressor.m b/first version/GetStimRegressor.m
new file mode 100644
index 0000000..ba2d9d5
--- /dev/null
+++ b/first version/GetStimRegressor.m	
@@ -0,0 +1,186 @@
+%%
+% all photoStates:
+% 0 = all black; 1 = black/white; 2 = white/black; 3 = all white; 4 = all gray;
+% 5 = gray/black; 6 = white/gray; 7 = black/gray; 8 = gray/white.
+% 10 = forward grating (very slow, more for calibration)
+% 11 = rightward grating
+% 12 = leftward grating
+
+%  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15
+% 00 01 02 03 10 11 12 13 20 21 22 23 30 31 32 33
+% transition e.g.:
+% photostate: 2     3     1     3     3     0     0     1     1     0     3     2     0     2     2     1
+% phototrans: 6    11    13     7    15    12     0     1     5     4     3    14     8     2    10     9
+
+%%
+function [regressors, names] = GetStimRegressor(stim,fishset)
+
+fpsec = 1.97; % should import from data...
+
+%% generate GCaMP6f kernel
+% GCaMP6f: decay half-time: 400±41; peak: 80±35 (ms)
+halftime = 0.4;
+delay = 0.08;
+% GCaMP6s: 1796±73, 480±24 (ms)
+% halftime = 1.8;
+% delay = 0.48;
+
+% the 2014-15 Janelia dataset used nucleus localized GCaMP6f...
+
+T = 8; % kernel time length in sec
+
+tlen=size(stim,2);
+t=0:0.05:T; % sec
+gc6=exp(-(t-(T/2+delay))/halftime);
+gc6(t<(T/2+delay))=0;
+t_im = 0:1/fpsec:1/fpsec*(tlen-1);
+t_gc6=0:0.05:tlen; % sec
+
+%% 
+if fishset == 1,
+    States = [0,1,2,3];
+    names = {'black','phototaxis left','phototaxis right','white',...
+        'right on','left on','right off','left off'};
+elseif fishset == 2,
+    States = [0,1,2,3,4,10,11,12];
+    names = {'black','phototaxis left','phototaxis right','white','grey','OMR forward','OMR left','OMR right',...
+        'left PT&OMR','right PT&OMR'};
+end
+tlen=length(stim);
+impulse = 6; % in frames % arbiturary at this point
+
+%% single photoStates
+numPS = length(States);
+stimPS_on = zeros(numPS,tlen); % PS = photoState
+% stimPS_start = zeros(numPS,tlen); % binary, ~ impulse1
+% stimPS_stop = zeros(numPS,tlen);
+for i = 1:numPS,
+    if ~isempty(find(stim==States(i),1)),
+        stimPS_on(i,:) = (stim==States(i));
+        
+%         diff_A = [stimPS_on(i,1) diff(stimPS_on(i,:))];
+%         %     if i==8, diff_A(end)=-1; end % manual correction: series always ends with PS=8
+%         ind_start_A = find(diff_A==1);
+%         ind_stop_A = find(diff_A==-1);
+%         
+%         for k = 1:length(ind_start_A),
+%             if (ind_start_A(k)+impulse<tlen) % 1*fcSampleRate = samples per 1 sec
+%                 stimPS_start(i,ind_start_A(k):(ind_start_A(k)+impulse)) = 1;
+%             else
+%                 stimPS_start(i,ind_start_A(k):end) = 1;
+%             end
+%         end
+%         for k = 1:length(ind_stop_A),
+%             if (ind_stop_A(k)+impulse<tlen)
+%                 stimPS_stop(i,ind_stop_A(k):(ind_stop_A(k)+impulse)) = 1;
+%             else
+%                 stimPS_stop(i,ind_stop_A(k):end) = 1;
+%             end
+%         end
+    end
+end
+
+%% Combos
+if fishset == 1,    
+    %% PT combos   
+    % right on: 2 3
+    % left on:  1 3
+    % right off: 0 1
+    % left off:  0 2
+    H = {[2 3],[1 3],[0 1],[0 2]};
+    numCB1 = 4;
+    stimCB_on = zeros(numCB1, tlen);
+    for i = 1:numCB1,
+        for j = 1:length(H{i});
+            ix = H{i}(j)+1;
+            ix2 = find(stimPS_on(ix,:));
+            if ~isempty(ix2),
+                stimCB_on(i,ix2) = 1;
+            end
+        end
+    end
+%     stimCB_start = zeros(numCB1, tlen);
+%     for i = 1:numCB1,
+%         for j = 1:length(H{i});
+%             ix = H{i}(j)+1;
+%             ix2 = find(stimPS_start(ix,:));
+%             if ~isempty(ix2),
+%                 stimCB_start(i,ix2) = 1;
+%             end
+%         end
+%     end    
+    
+elseif fishset == 2,  
+    %% PT-OMR combos
+    % left PT/OMR:  1 11
+    % right PT/OMR: 2 12
+    
+    H = {[1 11],[2 12]};
+    numCB1 = 2;
+    stimCB_on = zeros(numCB1, tlen);
+    for i = 1:numCB1,
+        for j = 1:length(H{i});
+            temp = (stim==H{i}(j));
+            ix = find(temp);
+            if ~isempty(ix),
+                stimCB_on(i,ix) = 1;
+            end
+        end
+    end
+    
+end
+
+%% pool all into cell array
+regressor_0={
+    stimPS_on;
+    stimCB_on;
+    };
+nRegType = length(regressor_0);
+% % name_array = {'stimPS_on','stimPS_start','stimPS_stop','stimTS','stimCB_on',...
+% %     'stimCB_start','stimCB'};
+
+% segment length and round off, for shuffled control
+segLength = floor(tlen/80);
+s4_ = tlen-mod(tlen,segLength);
+
+% initialize/preallocate struct
+n = nRegType*numPS;
+regressors(n).name = [];
+regressors(n).im = [];
+regressors(n).ctrl = [];
+
+%% feed all regressors into struct 'regressor_stim'
+idx = 0;
+for j=1:nRegType, %run_StimRegType_subset,
+    len = size(regressor_0{j},1);
+    for i=1:len, %run_PhotoState_subset,
+        idx = idx + 1;
+        reg = regressor_0{j}(i,:);
+        %         idx = (j-1)*nStimRegType + i;
+        
+        regressors(idx).name = names{idx};
+% %         regressor_stim(idx).name = [name_array{j} '_' num2str(i)];
+        % generate regressor in imaging time
+        regressors(idx).im = gen_reg_im(gc6, t_gc6, t_im, reg);
+        
+        % generate shuffled regressor
+        reg_ = reg(1:s4_);
+        reg2D = reshape(reg_, segLength, []);
+        indices = randperm(size(reg2D,2)); % reshuffle by indexing
+        shffreg = reg2D(:, indices);
+        shffreg = reshape(shffreg,1,[]);
+        temp = reg;
+        temp(1:s4_) = shffreg;
+        shffreg = temp;
+        regressors(idx).ctrl = gen_reg_im(gc6, t_gc6, t_im, shffreg);
+    end
+end
+end
+
+function reg_im = gen_reg_im(gc6, t_gc6, t_im, reg)
+
+temp1=interp1(t_im,reg,t_gc6,'linear','extrap');
+temp2=conv(temp1,gc6,'same');
+reg_im = interp1(t_gc6,temp2,t_im,'linear','extrap');
+
+end
diff --git a/Load_LSh_step1_cleanup.m b/first version/Load_LSh_step1_cleanup.m
similarity index 100%
rename from Load_LSh_step1_cleanup.m
rename to first version/Load_LSh_step1_cleanup.m
diff --git a/Load_LSh_step2_detrend.m b/first version/Load_LSh_step2_detrend.m
similarity index 100%
rename from Load_LSh_step2_detrend.m
rename to first version/Load_LSh_step2_detrend.m
diff --git a/Load_LSh_step3_align.m b/first version/Load_LSh_step3_align.m
similarity index 100%
rename from Load_LSh_step3_align.m
rename to first version/Load_LSh_step3_align.m
diff --git a/Load_LSh_step4_loadGUI.m b/first version/Load_LSh_step4_loadGUI.m
similarity index 100%
rename from Load_LSh_step4_loadGUI.m
rename to first version/Load_LSh_step4_loadGUI.m
diff --git a/Modeling001.m b/first version/Modeling001.m
similarity index 100%
rename from Modeling001.m
rename to first version/Modeling001.m
diff --git a/PlotBy16StimsFromGUI.m b/first version/PlotBy16StimsFromGUI.m
similarity index 100%
rename from PlotBy16StimsFromGUI.m
rename to first version/PlotBy16StimsFromGUI.m
diff --git a/Plotting.m b/first version/Plotting.m
similarity index 100%
rename from Plotting.m
rename to first version/Plotting.m
diff --git a/Plotting_withinM.m b/first version/Plotting_withinM.m
similarity index 100%
rename from Plotting_withinM.m
rename to first version/Plotting_withinM.m
diff --git a/PoolClustersFromAllFish.m b/first version/PoolClustersFromAllFish.m
similarity index 100%
rename from PoolClustersFromAllFish.m
rename to first version/PoolClustersFromAllFish.m
diff --git a/SortMbystim.m b/first version/SortMbystim.m
similarity index 100%
rename from SortMbystim.m
rename to first version/SortMbystim.m
diff --git a/first version/imNormalize99.m b/first version/imNormalize99.m
new file mode 100644
index 0000000..b35ac8e
--- /dev/null
+++ b/first version/imNormalize99.m	
@@ -0,0 +1,11 @@
+function out=imNormalize99(im)
+
+im=double(im);
+temp=sort(im(:),'descend');
+th1=temp(round(length(im(:))/100));
+th2=min(im(:));
+
+out=(im-th2)/(th1-th2);
+out(out>1)=1;
+
+
diff --git a/first version/push_cIX_gIX.m b/first version/push_cIX_gIX.m
new file mode 100644
index 0000000..0663718
--- /dev/null
+++ b/first version/push_cIX_gIX.m	
@@ -0,0 +1,10 @@
+function push_cIX_gIX(hfig,cIX,gIX,numK)
+% push cIX, gIX to hfig
+if ~exist('numK','var')
+    numK=max(gIX);
+end
+
+var_script={hfig,cIX,gIX,numK};
+GUI_FishExplorer('_place_holder_',0,0,0,1,'push_cIX_gIX',var_script);
+
+end
\ No newline at end of file
diff --git a/first version/read_LSstack_fast_float.m b/first version/read_LSstack_fast_float.m
new file mode 100644
index 0000000..b17963a
--- /dev/null
+++ b/first version/read_LSstack_fast_float.m	
@@ -0,0 +1,15 @@
+function [stack,dim]=read_LSstack_fast_float(fpath,dim)
+%% reads whole stack
+
+info=dir(fpath);
+if (isempty(info))
+    error('File Does not exist');
+end
+
+
+stack=read_LSstack_fast_float_mex64(fpath,int32(dim(1:2)));
+
+size(stack)
+
+
+stack=reshape(stack,dim);
diff --git a/fishpic.jpg b/fishpic.jpg
new file mode 100644
index 0000000..ef40e83
Binary files /dev/null and b/fishpic.jpg differ
diff --git a/read_LSstack_fast_float_mex64.mexw64 b/read_LSstack_fast_float_mex64.mexw64
new file mode 100644
index 0000000..d5df9b0
Binary files /dev/null and b/read_LSstack_fast_float_mex64.mexw64 differ