From 6fe24e422fc60700b124b67986bd8929b86bb070 Mon Sep 17 00:00:00 2001 From: Sebastian Mackowiak Date: Tue, 6 Sep 2011 17:05:32 +0200 Subject: [PATCH] version 0.0.1 --- miRDeep2.pl | 8 ++++---- prepare_signature.pl | 33 +++++++++++++++++++++++++++++++-- repath_pdfs.pl | 25 +++++++++++++++++++++++++ 3 files changed, 60 insertions(+), 6 deletions(-) create mode 100755 repath_pdfs.pl diff --git a/miRDeep2.pl b/miRDeep2.pl index 8c2a370..e18f97e 100755 --- a/miRDeep2.pl +++ b/miRDeep2.pl @@ -805,15 +805,15 @@ sub output_results{ if($file_mature_ref_this_species !~ /none/i){ if($options{'q'}){ - $line="make_html.pl -f $dir/output.mrd -k $dir_tmp/$file_mature_ref_this_species -p $dir_tmp/precursors.coords -s $dir/survey.csv -c -e -q $options{'q'} -x $xopt -r ${scripts}Rfam_for_miRDeep.fa -g $sc -y $time $sort_by_sample"; + $line="make_html.pl -f $dir/output.mrd -k $dir_tmp/$file_mature_ref_this_species -p $dir_tmp/precursors.coords -s $dir/survey.csv -c -e -q $options{'q'} -x $xopt -r ${scripts}Rfam_for_miRDeep.fa -v $sc -y $time $sort_by_sample"; }else{ - $line="make_html.pl -f $dir/output.mrd -k $dir_tmp/$file_mature_ref_this_species -p $dir_tmp/precursors.coords -s $dir/survey.csv -c -e -r ${scripts}Rfam_for_miRDeep.fa -g $sc -y $time $sort_by_sample"; + $line="make_html.pl -f $dir/output.mrd -k $dir_tmp/$file_mature_ref_this_species -p $dir_tmp/precursors.coords -s $dir/survey.csv -c -e -r ${scripts}Rfam_for_miRDeep.fa -v $sc -y $time $sort_by_sample"; } }else{ if($options{'q'}){ - $line="make_html.pl -f $dir/output.mrd -p $dir_tmp/precursors.coords -s $dir/survey.csv -c -e -q $options{'q'} -x $xopt -r ${scripts}Rfam_for_miRDeep.fa -g $sc -y $time $sort_by_sample"; + $line="make_html.pl -f $dir/output.mrd -p $dir_tmp/precursors.coords -s $dir/survey.csv -c -e -q $options{'q'} -x $xopt -r ${scripts}Rfam_for_miRDeep.fa -v $sc -y $time $sort_by_sample"; }else{ - $line="make_html.pl -f $dir/output.mrd -p $dir_tmp/precursors.coords -g $sc -s $dir/survey.csv -c -e -r ${scripts}Rfam_for_miRDeep.fa -y $time $sort_by_sample"; + $line="make_html.pl -f $dir/output.mrd -p $dir_tmp/precursors.coords -v $sc -s $dir/survey.csv -c -e -r ${scripts}Rfam_for_miRDeep.fa -y $time $sort_by_sample"; } } diff --git a/prepare_signature.pl b/prepare_signature.pl index b774d9d..0f924fd 100755 --- a/prepare_signature.pl +++ b/prepare_signature.pl @@ -63,13 +63,20 @@ # system("cat $dir/reads_vs_precursors.arf $dir/mature_vs_precursors.arf > $dir/signature_unsorted.arf"); #Sortarf("$dir/signature_unsorted.arf"); - system("sort -k6 $dir/signature_unsorted.arf > $outfile"); + presort("$dir/signature_unsorted.arf"); + system("sort -nk1 $dir/signature_unsorted.arf.tmp > $dir/signature_unsorted.arf.tmp2"); + system("cut -f2-14 $dir/signature_unsorted.arf.tmp2 > $outfile"); +# system("sort -V -k6 $dir/signature_unsorted.arf > $outfile"); + }else{ if($options{b}){print STDERR "sorting rows\n";} #Sortarf("$dir/reads_vs_precursors.arf"); - system("sort -k6 $dir/reads_vs_precursors.arf > $outfile"); + presort("$dir/reads_vs_precursors.arf"); + system("sort -nk1 $dir/reads_vs_precursors.arf.tmp > $dir/reads_vs_precursors.arf.tmp2"); + system("cut -f2-f14 $dir/reads_vs_precursors.arf.tmp2 > $outfile"); +# system("sort -V -k6 $dir/reads_vs_precursors.arf > $outfile"); } @@ -88,6 +95,28 @@ ############################################################ ############################################################ +sub presort{ + my $file=shift; + open IK,"$file" or die "no arf file given\n"; + open IKT,">$file.tmp" or die "tmp file could not be opened\n"; + + my %index=(); + my $count=0; + my @l; + + while(){ + @l=split(); + if(not $index{$l[5]}){ + $count++; + $index{$l[5]}=$count; + } + print IKT "$index{$l[5]}\t$_"; + } + close IK; + close IKT; +} + + sub Sortarf { my $file=shift; open IN,"<$file" or die "FILE $file not found $!\n"; diff --git a/repath_pdfs.pl b/repath_pdfs.pl new file mode 100755 index 0000000..01ae636 --- /dev/null +++ b/repath_pdfs.pl @@ -0,0 +1,25 @@ +#!/usr/bin/perl + +use strict; +use File::Basename; +use Cwd; + +my $cwd = cwd; +$cwd .="/"; + +my $pdfsdir = $ARGV[1] or die "Usage: $0 html_file pdfs_dir > outfile.html\n"; + +open IN,"$ARGV[0]" or die "No html input file given\n"; +my ( $name0, $path0, $extension0 ) = fileparse ( $ARGV[0], '\..*' ); + + +open OUT,">${name0}_repath.html" or die "Could not create file ${name0}_repath.html\n"; + +while(){ + if(/a href="file:\/\/(\/\S+)$pdfsdir/){ + $_ =~ s/$1/$cwd/; + } + print OUT; +} +close IN; +close OUT;