diff htseq.pl @ 4:ebd59bc6855c draft

Uploaded
author fcaramia
date Wed, 12 Sep 2012 23:45:33 -0400
parents
children 6324eefd9e91
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/htseq.pl	Wed Sep 12 23:45:33 2012 -0400
@@ -0,0 +1,107 @@
+#!/usr/bin/perl
+
+use strict;
+use warnings;
+use Getopt::Std;
+use File::Basename;
+$| = 1;
+
+# Grab and set all options
+my %OPTIONS = (a => 0, i => "gene_id", m => "intersection-nonempty", s => "no", t => "exon");
+getopts('a:cg:i:m:o:r:s:t:', \%OPTIONS);
+
+die qq(
+Usage:   HTSeq.pl [OPTIONS] Group1=sample1=<SAM/BAM file> [Group1=sample2=<SAM/BAM file> ... Group2=sampleN=<SAM/BAM file> ...]
+
+OPTIONS:	-a	STR	skip all reads with alignment quality lower than the given minimum value (default: $OPTIONS{a})
+			-c		reduce the matrix by removing any feature with no counts
+			-g	STR	the features file in the GFF/GTF format
+			-i	STR	GFF attribute to be used as feature ID (default: $OPTIONS{i})
+			-m	STR	mode to handle reads overlapping more than one feature. Possible values for <mode> are union, intersection-strict and intersection-nonempty (default: $OPTIONS{m})
+			-o	STR	output file name for expression matrix
+			-r	STR	the name of the output report
+			-s	STR	whether the data is from a strand-specific assay (default: $OPTIONS{s})
+			-t	STR	feature type (3rd column in GFF file) to be used, all features of other type are ignored (default, suitable for RNA-Seq and Ensembl GTF files: $OPTIONS{t})
+
+) if(@ARGV == 0);
+
+my $sam_out;
+my @counts;
+my @features;
+my %report;
+my @samplenames;
+my $current_group;
+my @groups;
+my @files;
+my $groupcount = 0;
+my %grouphash;
+
+foreach my $input (@ARGV) {
+	my ($group, $sample, $input) = split "::", $input;
+	if(! defined $grouphash{$group}) {
+		$groupcount++;
+		$grouphash{$group} = "G${groupcount}:$group";
+	}
+	push @groups, $grouphash{$group};
+	push @samplenames, $sample;
+	push @files, $input;
+}
+
+for(my $index = 0; $index <= $#files; $index++) {
+	my $input_file = $files[$index];
+	my $sample = $samplenames[$index];
+	
+	# run htseq
+	my @htseq;
+	my $COMM;
+	my $file_type = `file $input_file`;
+	if(grep /text$/, $file_type ) {
+		$COMM = "htseq-count -q -m $OPTIONS{m} -s $OPTIONS{s} -a $OPTIONS{a} -t $OPTIONS{t} -i $OPTIONS{i} $input_file $OPTIONS{g}";
+		@htseq = `$COMM`;
+	} else {
+		$COMM = "samtools view $input_file | htseq-count -q -m $OPTIONS{m} -s $OPTIONS{s} -a $OPTIONS{a} -t $OPTIONS{t} -i $OPTIONS{i} - $OPTIONS{g}";
+		@htseq = `$COMM`;
+	}
+	
+	my $row = 0;
+	$report{$sample} = "Command Used: $COMM\n";
+
+	for(my $row = 0; $row <= $#htseq; $row++) {
+		# store the report is an hash
+		if(grep /^no_feature|^ambiguous|^too_low_aQual|^not_aligned|^alignment_not_unique/, $htseq[$row]) {
+			$report{$sample} .= $htseq[$row];
+		} else {
+			# store the counts in a matrix
+			chomp $htseq[$row];
+			my ($feature, $value) = split "\t", $htseq[$row];
+			$counts[$row][$index] = $value;
+			if($input_file eq $files[0]) {
+				push @features, $feature;
+			}
+		}
+	}
+}
+
+# print the matrix
+open(MATRIX, ">$OPTIONS{o}") || die "Could Not Create Output File $OPTIONS{o}!\n";
+print MATRIX "#\t".join("\t", @groups)."\n";
+print MATRIX "#Feature\t".join("\t", @samplenames)."\n";
+for(my $row = 0; $row <= $#features; $row++) {
+	if(defined $OPTIONS{c}) {
+		my $rowsum = 0;
+		$rowsum += $_ foreach @{ $counts[$row] };
+		if(!$rowsum) {
+			next;
+		}
+	}
+	print MATRIX "$features[$row]\t".join("\t", @{ $counts[$row] })."\n";
+}
+close(MATRIX);
+
+# print the report
+open(REPORT, ">$OPTIONS{r}") || die "Could Not Create Output File $OPTIONS{r}!\n";
+print REPORT "$groups[$_]:$samplenames[$_]\n$report{$samplenames[$_]}\n" foreach (0..$#samplenames);
+close(REPORT);
+
+
+