From: elserj <elserj@localhost>
Date: Fri, 21 Apr 2017 22:33:09 +0000 (+0000)
Subject: Add script that will look for known isoform patterns and create a new fasta file... 
X-Git-Url: http://gitweb.planteome.org/?a=commitdiff_plain;h=b65f9ad425a76039084f8747492e91fda7331fe9;p=old-jaiswallab-svn%2F.git

Add script that will look for known isoform patterns and create a new fasta file without them

svn path=/; revision=655
---

diff --git a/interactome_scripts/fasta_header_and_isoform_fix.pl b/interactome_scripts/fasta_header_and_isoform_fix.pl
new file mode 100755
index 0000000..e9887b5
--- /dev/null
+++ b/interactome_scripts/fasta_header_and_isoform_fix.pl
@@ -0,0 +1,79 @@
+#!/usr/bin/perl
+
+use strict;
+use warnings;
+
+if(-e "$ENV{HOME}/scripts/jaiswallab/interactome_scripts/find_species.pl") {
+	require "$ENV{HOME}/scripts/jaiswallab/interactome_scripts/find_species.pl";
+}elsif(-e "$ENV{HOME}/jaiswallab_svn/interactome_scripts/find_species.pl") {
+	require "$ENV{HOME}/jaiswallab_svn/interactome_scripts/find_species.pl";
+}
+
+my @files;
+
+if($#ARGV != 1) {
+	print "Usage: fasta_header_fix.pl input_file output_file\n";
+	exit;
+}
+
+my $input_file = $ARGV[0];
+my $output_file = $ARGV[1];
+	
+open(INFILE, $input_file)  || die "Error: File $input_file can not be opened";
+open(OUTFILE, ">$output_file") || die "Error: File $output_file can not be opened";
+	
+my $species = find_species($input_file);
+
+my  $is_isoform = 0;
+
+while(<INFILE>) {
+	my $line = $_;
+	chomp $line;
+	if ($line =~ /^>/) {
+		my $gene = find_gene($line,$species);
+		$is_isoform = check_isoform($gene);
+		next if($is_isoform);
+		$line = ">" . "$gene";
+	}
+	next if ($is_isoform);
+	print OUTFILE "$line\n";
+}
+
+close(INFILE);
+close(OUTFILE);
+
+#subroutine to check if current gene is an isoform
+sub check_isoform {
+		my $gene = $_[0];
+
+		#Maize gene ids are different
+		if($species eq "Zea_mays") {
+				print "in here\t$gene\n";
+				if ($gene =~ /\_T0[2-9]$/ || $gene =~ /\_T1[0-9]$/) {
+						return "1";
+				}
+		}
+		
+		# So are theobroma
+		if($species eq "Theobroma_cacao") {
+				if ($gene =~ /t[2-9]$/ || $gene =~ /t1[0-9]$/) {
+						return "1";
+				}
+		}
+		# Jatropha
+		if($species eq "Jatropha_curcas") {
+				if ($gene =~ /\.20$/ || $gene =~ /\.30/ || $gene =~ /\.40/ || $gene =~ /\.50/ || $gene =~ /\.60/ || $gene =~ /\.70/ || $gene =~ /\.80/ || $gene =~ /\.90/) {
+						return "1";
+				}
+		}
+		#all other cases
+		if($species ne "Jatropha_curcas") { #Jatropha gene ids are .10, .20, .30, etc...
+				if ($gene =~ /\.[2-9]$/ || $gene =~ /\.1[0-9]$/ || $gene =~ /\.2[0-9]$/ || $gene =~ /\.3[0-9]$/ || $gene =~ /\.4[0-9]$/ || $gene =~ /\.5[0-9]$/ || $gene =~ /\.6[0-9]$/ || $gene =~ /\.7[0-9]$/ || $gene =~ /\.8[0-9]$/ || $gene =~ /\.9[0-9]$/) {
+						return "1";
+				}
+		}
+		
+		
+		# if not found above, assume not an isoform
+		return "0";
+}