interval_method option for statistical summaries

Changes

@@ -1,7 +1,11 @@
 Revision history for Perl extension Bio::Graphics.
 
+2.31 Tue Sep 25 22:39:43 EDT 2012
+   - Fix infinite loop when drawing wiggle_xyplots with z-score scaling
+     over regions that do not vary much.
+
 2.30
-    - Added the glyph for FlyBase's "stacked wiggle" plot (fb_shmiggle.pl) as well as a utility script for preparing the data (index_cov_files.pl).
+    - Added the glyph for FlyBase's "stacked wiggle" plot (fb_shmiggle.pm) as well as a utility script for preparing the data (index_cov_files.pl).
 
 2.29
     - Fixes for Bio::Graphics::Wiggle to prevent crash when the "bottom" keystyle selected.

MANIFEST

@@ -127,8 +127,6 @@ scripts/render_msa.pl
 scripts/search_overview.pl
 t/BioGraphics.t
 t/data/feature_data.txt
-t/data/t1.gif
-t/data/t1.png
 t/data/t1/version1.gif
 t/data/t1/version1.png
 t/data/t1/version10.png
@@ -149,8 +147,6 @@ t/data/t1/version7.png
 t/data/t1/version8.png
 t/data/t1/version9.gif
 t/data/t1/version9.png
-t/data/t2.gif
-t/data/t2.png
 t/data/t2/version1.gif
 t/data/t2/version1.png
 t/data/t2/version10.png
@@ -178,8 +174,6 @@ t/data/t2/version6.png
 t/data/t2/version7.png
 t/data/t2/version8.png
 t/data/t2/version9.png
-t/data/t3.gif
-t/data/t3.png
 t/data/t3/version1.gif
 t/data/t3/version1.png
 t/data/t3/version10.gif

lib/Bio/Graphics.pm

@@ -2,7 +2,7 @@ package Bio::Graphics;
 
 use strict;
 use Bio::Graphics::Panel;
-our $VERSION = '2.29';
+our $VERSION = '2.31';
 
 1;
 

lib/Bio/Graphics/Glyph/hybrid_plot.pm

@@ -70,7 +70,7 @@ sub draw {
  my ($left,$top,$right,$bottom) = $self->calculate_boundaries($dx,$dy);
  my $height   = $bottom - $top;
  my $feature  = $self->feature;
- my $set_flip = $self->option('flip_sign') | 0;
+ my $set_flip = $self->option('flip_sign') || 0;
 
  #Draw individual features for reads (unlike wiggle features reads will have scores)
  my $t_id = $feature->method;
@@ -106,7 +106,23 @@ sub draw {
 					-end    => $self->panel->end,
 					-type   => 'summary');
 	 my $stats = $summary->statistical_summary($self->width);
-	 my @vals  = map {$_->{validCount} ? $_->{sumData}/$_->{validCount}*$flip:0} @$stats;
+	 my $interval_method = $self->option('interval_method') || 'mean';
+	 my @vals;
+	 if ($interval_method eq 'mean') {
+		@vals  = map {$_->{validCount} ? $_->{sumData}/$_->{validCount} * $flip : undef} @$stats;
+	 }
+	 elsif ($interval_method eq 'sum') {
+		@vals  = map {$_->{validCount} ? $_->{sumData} * $flip : undef} @$stats;
+	 }
+	 elsif ($interval_method eq 'min') {
+		@vals  = map {$_->{validCount} ? $_->{minVal} * $flip : undef} @$stats;
+	 }
+	 elsif ($interval_method eq 'max') {
+		@vals  = map {$_->{validCount} ? $_->{maxVal} * $flip : undef} @$stats;
+	 }
+	 else {
+		warn "unrecognized interval method $interval_method!";
+	 }
 	 $self->_draw_coverage($summary,\@vals,@_);
      }
  }
@@ -129,6 +145,7 @@ sub minmax {
     my $parts  = shift;
 
     my $autoscale  = $self->option('autoscale') || 'local';
+    my $set_flip = $self->option('flip_sign') || 0;
 
     my $min_score  = $self->min_score  unless $autoscale eq 'z_score';
     my $max_score  = $self->max_score  unless $autoscale eq 'z_score';
@@ -160,7 +177,8 @@ sub minmax {
 	$stdev += $d**2;
     }
     $stdev = sqrt($stdev);
-
+    $min = $max * -1 if ($set_flip);
+
     $min_score = $min if $do_min;
     $max_score = $max if $do_max;
     return $self->sanity_check($min_score,$max_score,$mean,$stdev);

lib/Bio/Graphics/Glyph/vista_plot.pm

@@ -203,7 +203,23 @@ sub draw_signal {
 	    $self->Bio::Graphics::Glyph::wiggle_whiskers::draw(@_);
 	} else {
 	    my $stats = $summary->statistical_summary($self->width);
-	    my @vals  = map {$_->{validCount} ? Bio::DB::BigWig::binMean($_) : 0} @$stats; 
+		my $interval_method = $self->option('interval_method') || 'mean';
+		my @vals;
+		if ($interval_method eq 'mean') {
+			@vals  = map {$_->{validCount} ? Bio::DB::BigWig::binMean($_) : undef} @$stats;
+		}
+		elsif ($interval_method eq 'sum') {
+			@vals  = map {$_->{validCount} ? $_->{sumData} : undef} @$stats;
+		}
+		elsif ($interval_method eq 'min') {
+			@vals  = map {$_->{validCount} ? $_->{minVal} : undef} @$stats;
+		}
+		elsif ($interval_method eq 'max') {
+			@vals  = map {$_->{validCount} ? $_->{maxVal} : undef} @$stats;
+		}
+		else {
+			warn "unrecognized interval method $interval_method!";
+		}
 	    $self->bigwig_summary($summary);
 	    if ($signal_type eq 'density') {
 		$self->Bio::Graphics::Glyph::wiggle_density::_draw_coverage($summary,\@vals,@_);

lib/Bio/Graphics/Glyph/wiggle_data.pm

@@ -241,7 +241,23 @@ sub create_parts_from_summary {
     my $self = shift;
     my ($stats,$start,$end) = @_;
     $stats ||= [];
-    my @vals  = map {$_->{validCount} ? $_->{sumData}/$_->{validCount}:0} @$stats;
+    my $interval_method = $self->option('interval_method') || 'mean';
+    my @vals;
+    if ($interval_method eq 'mean') {
+    	@vals  = map {$_->{validCount} ? $_->{sumData}/$_->{validCount} : undef} @$stats;
+    }
+    elsif ($interval_method eq 'sum') {
+    	@vals  = map {$_->{validCount} ? $_->{sumData} : undef} @$stats;
+    }
+    elsif ($interval_method eq 'min') {
+    	@vals  = map {$_->{validCount} ? $_->{minVal} : undef} @$stats;
+    }
+    elsif ($interval_method eq 'max') {
+    	@vals  = map {$_->{validCount} ? $_->{maxVal} : undef} @$stats;
+    }
+    else {
+    	warn "unrecognized interval method $interval_method!";
+    }
     return \@vals;
 }
 
@@ -395,7 +411,23 @@ sub draw_statistical_summary {
     my $feature = shift;
     my $stats = $feature->statistical_summary($self->width);
     $stats   ||= [];
-    my @vals  = map {$_->{validCount} ? $_->{sumData}/$_->{validCount}:0} @$stats;
+    my $interval_method = $self->option('interval_method') || 'mean';
+    my @vals;
+    if ($interval_method eq 'mean') {
+    	@vals  = map {$_->{validCount} ? $_->{sumData}/$_->{validCount} : undef} @$stats;
+    }
+    elsif ($interval_method eq 'sum') {
+    	@vals  = map {$_->{validCount} ? $_->{sumData} : undef} @$stats;
+    }
+    elsif ($interval_method eq 'min') {
+    	@vals  = map {$_->{validCount} ? $_->{minVal} : undef} @$stats;
+    }
+    elsif ($interval_method eq 'max') {
+    	@vals  = map {$_->{validCount} ? $_->{maxVal} : undef} @$stats;
+    }
+    else {
+    	warn "unrecognized interval method $interval_method!";
+    }
     return $self->_draw_coverage($feature,\@vals,@_);
 }
 
@@ -416,6 +448,7 @@ sub _draw_coverage {
 	my $s      = $start + $offset;
 	my $e      = $s+1;  # fill in gaps
 	my $v      = $array->[$offset/$bases_per_bin];
+	next unless defined $v; # skip missing values
 	push @parts,[$s,$s,$v];
     }
     $self->draw_plot(\@parts,@_);

lib/Bio/Graphics/Glyph/wiggle_density.pm

@@ -401,6 +401,7 @@ sub _draw_coverage {
 	my $s      = $start + $offset;
 	my $e      = $s+1;  # fill in gaps
 	my $v      = $array->[$offset/$bases_per_bin];
+	$v         = 0 unless defined $v; # don't want undefined values
 	push @parts,[$s,$s,$v];
     }
     $self->Bio::Graphics::Glyph::wiggle_density::draw_plot(\@parts,@_);

lib/Bio/Graphics/Glyph/wiggle_xyplot.pm

@@ -11,10 +11,12 @@ sub my_description {
     return <<END;
 This glyph draws quantitative data as an xyplot. It is designed to be
 used in conjunction with features in "wiggle" format as generated by
-Bio::Graphics::Wiggle, or base pair coverage data generated by the
-Bio::DB::Sam module.
+Bio::Graphics::Wiggle, base pair coverage data generated by the
+Bio::DB::Sam module, or interval statistical_summaries generated by 
+the Bio::DB::BigWig module.
 
-For this glyph to work, the feature must define one of the following tags:
+For this glyph to work, the feature must define one of the following tags
+or methods:
 
   wigfile -- a path to a Bio::Graphics::Wiggle file
 
@@ -23,7 +25,9 @@ For this glyph to work, the feature must define one of the following tags:
 
   coverage-- a simple comma-delimited string containing the quantitative values,
              assumed to be one value per pixel.
-
+
+  statistical_summary() -- a method for generating statistical information 
+             including validCount, maxVal, minVal, sumData, and sumSquares.
 END
 }
 
@@ -64,6 +68,13 @@ sub my_options {
 	    'scaled to the minimum and maximum values of the region currently on display.',
 	    'min_score and max_score override autoscaling if one or both are defined'
         ],
+    interval_method => [
+    	['mean', 'sum', 'min', 'max'],
+    	'mean',
+    	'When working with features that offer a statistical_summary() method,',
+    	'such as those from Bio::DB::BigWig, define the method for reporting',
+    	'scores within each interval.',
+    ],
     };
 }
 
@@ -147,8 +158,8 @@ sub draw_plot {
 	$scaled_min = int(($min_score-$mean)/$stdev + 0.5);
 	$scaled_max = int(($max_score-$mean)/$stdev + 0.5);
 	my $bound  = $self->z_score_bound;
-	$scaled_max = $bound  if $scaled_max > 0;
-	$scaled_min = -$bound if $scaled_min < 0;
+	$scaled_max = $bound  if $scaled_max >= 0;
+	$scaled_min = -$bound if $scaled_min <= 0;
     }
     elsif ($side) {
 	$scaled_min = int($min_score - 0.5);
@@ -396,6 +407,12 @@ the "wigdata" attribute. This can be an arrayref of quantitative data
 starting at feature start and ending at feature end, or the
 data string returned by Bio::Graphics::Wiggle->export_to_wif64($start,$end).
 
+The glyph also supports features which offer a statistical_summary()
+method, such as those from Bio::DB::BigWig. This method returns a hash
+of values, including validCount, maxVal, minVal, sumData, and sumSquares.
+For each interval, a statistical score is generated from these values.
+The mean, minimum, maximum, or sum of the values may be reported.
+
 =head2 OPTIONS
 
 In addition to all the xyplot glyph options, the following options are
@@ -439,6 +456,12 @@ recognized:
    neg_color   color        When drawing bicolor plots, the fill color to use for values
                               that are below the pivot point.
 
+   interval_method   method  When working with bigWig features that use the
+                             statistical_summary() method, define the method
+                             for reporting one or more scores within each
+                             interval. Options include "mean", "sum", "min",
+                             or "max". Default is "mean".
+
 =head2 SPECIAL FEATURE TAGS
 
 The glyph expects one or more of the following tags (attributes) in