Parse BES-flaxESTs

<source lang=perl>

1. !/usr/bin/perl -w

1. Libraries

use Bio::SearchIO; use Parallel::ForkManager;

1. Variables

my ($i) = $_;

1. Input prompt

if (! $ARGV[0]) { print "What is the BLAST file to parse? "; chomp ($inFile = <STDIN>); } else { $inFile = $ARGV[0]; }

1. Multithreading

my $pm = new Parallel::ForkManager(16);

1. Some counting

my $c_entries = `grep -c "Sequences producing significant alignments" $inFile`; chomp $c_entries;

1. Making the outfile

open (GFF, ">".$inFile.".csv"); print GFF "Query\tHitName\tHitDesc\tHSP_rank\t\%ID\teValue\tHSP_length\n"; close GFF;

1. Parsing Function

$report = new Bio::SearchIO(

        		-file=>"$inFile",

-format => "blast");

$i = 0; # Go through BLAST reports one by one while($result = $report->next_result) {

$i++; $pm->start and next;

# Go through each each matching sequence while($hit = $result->next_hit) {

# Go through each each HSP for this sequence while ($hsp = $hit->next_hsp) {

# Capture the results my $accesssion = $result->query_accession; my $hitname = $hit->name;

				my $HitDesc    = $hit->description;
				my $rank       = $hsp->rank;  

my $identity = $hsp->percent_identity; my $evalue = $hsp->evalue; my $hsp_length = $hsp->hsp_length;

#Conditions next if $hsp->percent_identity < 80 ; next if $hsp->hsp_length < 80 ; $identity = sprintf ("%.2f", $identity);

# Writing into files open (GFF, ">>".$inFile.".csv"); print GFF "$accesssion\t$hitname\t$HitDesc\t$rank\t$identity\t$evalue\t$hsp_length\n"; close GFF;

#Decorating the log my $percent = ($i/$c_entries)*100; $percent = sprintf("%.2f", $percent);

print "\n\tCompleted [$i of $c_entries] | $percent% ";

} }

$pm->finish; } $pm->wait_all_children;

1. END

</perl>