diff --git a/schemas/+nc/GpfaCovExpl.m b/schemas/+nc/GpfaCovExpl.m
deleted file mode 100644
index 2942d99..0000000
--- a/schemas/+nc/GpfaCovExpl.m
+++ /dev/null
@@ -1,112 +0,0 @@
-%{
-nc.GpfaCovExpl (computed) # Covariance explained by GPFA model
-
--> nc.GpfaModelSet
-latent_dim           : tinyint unsigned  # number of latent dimensions
-by_trial             : boolean           # use spike counts for entire trial
----
-avg_var_expl_train          : double     # avg var expl on train set model based
-avg_var_expl_test           : double     # avg var expl on test set model based
-corr_resid_train            : mediumblob # residual cov for train set (avg over CV runs)
-corr_resid_test             : mediumblob # residual cov for test set (avg over CV runs)
-rmsd_corr_pred_train        : double     # RMS diff of offdiags for prediction and train set
-rmsd_corr_pred_test         : double     # RMS diff of offdiags for prediction and train set
-rmsd_corr_train_test        : double     # RMS diff of offdiags for prediction and train set
-%}
-
-classdef GpfaCovExpl < dj.Relvar & dj.AutoPopulate
-    properties(Constant)
-        table = dj.Table('nc.GpfaCovExpl');
-        popRel = nc.GpfaParams * nc.GpfaModelSet & 'kfold_cv > 1';
-    end
-
-    methods 
-        function self = GpfaCovExpl(varargin)
-            self.restrict(varargin{:})
-        end
-    end
-
-    methods (Access = protected)
-        function makeTuples(self, key)
-            Yt = fetch1(nc.GpfaModelSet & key, 'transformed_data');
-            par = fetch(nc.GpfaParams & key, '*');
-            coeff = @(C) C ./ sqrt(diag(C) * diag(C)');
-            for p = 0 : par.max_latent_dim
-                for byTrial = [false, true]
-                    vetrain = 0;
-                    vetest = 0;
-                    Qrtrain = 0;
-                    Qrtest = 0;
-                    rmsPredTrain = 0;
-                    rmsPredTest = 0;
-                    rmsTrainTest = 0;
-                    for k = 1 : par.kfold_cv
-                        modelKey = key;
-                        modelKey.latent_dim = p;
-                        modelKey.cv_run = k;
-                        [train, test, model] = fetch1(nc.GpfaModel & modelKey, 'train_set', 'test_set', 'model');
-                        model = GPFA(model);
-                        Ytrain = Yt(:, :, train);
-                        Ytest = Yt(:, :, test);
-
-                        % variance explained
-                        vetrain = vetrain + model.varExpl(Ytrain, byTrial) / par.kfold_cv;
-                        vetest = vetest + model.varExpl(Ytest, byTrial) / par.kfold_cv;
-
-                        % residual correlations
-                        Rtrain = model.residCov(Ytrain, byTrial);
-                        Qrtrain = Qrtrain + Rtrain;
-                        Rtest = model.residCov(Ytest, byTrial);
-                        Qrtest = Qrtest + Rtest;
-                        Qpred = coeff(model.C * model.C' + model.R);
-
-                        % RMS difference of predicted and observed correlations
-                        Qtrain = process(model, Ytrain, byTrial);
-                        Qtest = process(model, Ytest, byTrial);
-                        rmsPredTrain = rmsPredTrain + sqrt(mean(offdiag(Qpred - coeff(Qtrain)) .^ 2)) / par.kfold_cv;
-                        rmsPredTest = rmsPredTest + sqrt(mean(offdiag(Qpred - coeff(Qtest)) .^ 2)) / par.kfold_cv;
-                        rmsTrainTest = rmsTrainTest + sqrt(mean(offdiag(coeff(Qtrain) - coeff(Qtest)) .^ 2)) / par.kfold_cv;
-                    end
-
-                    % insert into database
-                    tuple = key;
-                    tuple.latent_dim = p;
-                    tuple.by_trial = byTrial;
-                    tuple.avg_var_expl_train = mean(vetrain);
-                    tuple.avg_var_expl_test = mean(vetest);
-                    tuple.corr_resid_train = coeff(Qrtrain);
-                    tuple.corr_resid_test = coeff(Qrtest);
-                    tuple.rmsd_corr_pred_train = rmsPredTrain;
-                    tuple.rmsd_corr_pred_test = rmsPredTest;
-                    tuple.rmsd_corr_train_test = rmsTrainTest;
-                    self.insert(tuple);
-
-                    % insert variance explained per cell
-                    unitIds = fetchn(nc.GpfaUnits & key, 'unit_id', 'ORDER BY unit_id');
-                    nUnits = numel(unitIds);
-                    for i = 1 : nUnits
-                        tuple = key;
-                        tuple.latent_dim = p;
-                        tuple.by_trial = byTrial;
-                        tuple.unit_id = unitIds(i);
-                        tuple.var_expl_train = vetrain(i);
-                        tuple.var_expl_test = vetest(i);
-                        insert(nc.GpfaVarExpl, tuple);
-                    end
-                end
-            end
-        end
-    end
-end
-
-
-function Q = process(model, Y, byTrial)
-
-Y0 = model.subtractMean(Y);
-if byTrial
-    Y0 = permute(sum(Y0, 2), [3 1 2]);
-else
-    Y0 = Y0(1 : end, :)';
-end
-Q = Y0' * Y0 / size(Y0, 1);
-end
diff --git a/schemas/+nc/GpfaVarExpl.m b/schemas/+nc/GpfaVarExpl.m
deleted file mode 100644
index 24b8282..0000000
--- a/schemas/+nc/GpfaVarExpl.m
+++ /dev/null
@@ -1,21 +0,0 @@
-%{
-nc.GpfaVarExpl (computed) # Variance explained for GPFA model
-
--> nc.GpfaUnits
--> nc.GpfaCovExpl
----
-var_expl_train          : double     # avg var expl on train set model based
-var_expl_test           : double     # avg var expl on test set model based
-%}
-
-classdef GpfaVarExpl < dj.Relvar
-    properties(Constant)
-        table = dj.Table('nc.GpfaVarExpl');
-    end
-
-    methods
-        function self = GpfaVarExpl(varargin)
-            self.restrict(varargin{:})
-        end
-    end
-end