Tgicl parse.pl

From RAJ INFO
Revision as of 13:44, 28 December 2009 by Raj (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search
#! /usr/bin/perl -w


# *************************************************************
# This Programme does the following in sequence
# -> Open a results folder of TGICL
# -> parse the contigs and singlets
# -> Output as multifasta of single file
# Author: Rajkumar (itc@rajkumar.in)

# *************************************************************

# Libraries
	use strict;
	use Bio::SeqIO;
	use Bio::Perl;

## Variables
	my $path         = "/home/raj/bio";
	my $resdir       = "$path/euca/unigene";

	my $singlets     = "$resdir/globulus_singlets.fa";
	my $contigs      = "$resdir/globulus_contigs.fa";
	my $unigenes     = "$resdir/globulus_unigenes.fa";
	my $contigs_temp = "$resdir/contigs_temp.fa";
	

	my @asmdir   = ("asm_1", "asm_2", "asm_3", "asm_4");

# Getting the grip
	if (-e "$singlets") {`rm  $singlets`;}
	if (-e "$contigs") {`rm  $contigs`;}
	if (-e "$unigenes") {`rm  $unigenes`;}
	if (-e "$contigs_temp") {`rm  $contigs_temp`;}

		
	foreach my $asm (@asmdir) {
		`cat $resdir/result/$asm/contigs >>$contigs_temp`;	
		`cat $resdir/result/$asm/singlets >>$singlets`;
	}

# Removing the contigs that have more than or equal to 5 EST pices
 
	my $in  = Bio::SeqIO->new(-file => "$contigs_temp", -format => 'Fasta');
	my $identifier;
	my $sequence;
	
	while ( my $seq = $in->next_seq() ) {
		$identifier = $seq->id;
  		$sequence = $seq->seq;

		
		my $contig_id = "$identifier";
		if ($contig_id =~ m/^(CL[0-9]+)Contig[0-9]+/g) { $contig_id = $1; }
		
		next if ($contig_id eq "CL1");
		next if ($contig_id eq "CL2");
		next if ($contig_id eq "CL3");
		next if ($contig_id eq "CL4");
		next if ($contig_id eq "CL5");
		next if ($contig_id eq "CL6");
		next if ($contig_id eq "CL7");
		next if ($contig_id eq "CL8");
		next if ($contig_id eq "CL9");
		next if ($contig_id eq "CL10");
		next if ($contig_id eq "CL11");
		next if ($contig_id eq "CL12");
		next if ($contig_id eq "CL13");
		next if ($contig_id eq "CL14");
		next if ($contig_id eq "CL15");
		next if ($contig_id eq "CL16");
		next if ($contig_id eq "CL17");
		next if ($contig_id eq "CL18");
		next if ($contig_id eq "CL19");
		next if ($contig_id eq "CL20");
		next if ($contig_id eq "CL21");
		next if ($contig_id eq "CL22");
		next if ($contig_id eq "CL23");
		next if ($contig_id eq "CL24");
		next if ($contig_id eq "CL25");
		next if ($contig_id eq "CL26");	
		next if ($contig_id eq "CL27");
		next if ($contig_id eq "CL28");		
		next if ($contig_id eq "CL29");

		my $fasta = ">$contig_id\n$sequence\n";

		open (H, ">>".$contigs);
		print H "$fasta";
		close H;


	}
	
	#Finally write the unigenes	
	`cat $contigs $singlets >$unigenes`;

# END