From: elserj <elserj@localhost>
Date: Wed, 4 May 2011 17:51:43 +0000 (+0000)
Subject: Add file to create a fasta protein file from a list of genes
X-Git-Url: http://gitweb.planteome.org/?a=commitdiff_plain;h=1825399d0864d0ab42b6bc647bbacb52fdf4970d;p=old-jaiswallab-svn%2F.git

Add file to create a fasta protein file from a list of genes

svn path=/; revision=92
---

diff --git a/interactome_scripts/make_fasta_from_gene_list.pl b/interactome_scripts/make_fasta_from_gene_list.pl
new file mode 100755
index 0000000..45be6a4
--- /dev/null
+++ b/interactome_scripts/make_fasta_from_gene_list.pl
@@ -0,0 +1,48 @@
+#!/usr/bin/perl
+
+use strict;
+use warnings;
+
+use lib "$ENV{HOME}/scripts/jaiswallab/interactome_scripts";
+
+use DbiFloret;
+
+if ($#ARGV != 2) {
+	print "usage: make_fasta_from_gene_list.pl species gene_list output_file\n";
+	exit;
+}
+
+my $species = $ARGV[0];
+
+my $gene_list_file = $ARGV[1];;
+
+my @genes;
+open(gene_file, $gene_list_file);
+while(<gene_file>) {
+	my $line = $_;
+	chomp $line;
+	push (@genes, $line);
+}
+close(gene_file);
+
+my $dbh = DbiFloret::dbconnect;
+
+my $safe_species = $dbh->quote_identifier($species);
+
+my $sth_get_sequence = $dbh->prepare("select sequence from $safe_species where gene_id like ?");
+
+open(outputfh, ">$ARGV[2]");
+
+foreach my $gene (@genes) {
+	$sth_get_sequence->execute($gene);
+	print outputfh ">$gene\n";
+	while (my @line = $sth_get_sequence->fetchrow_array) {
+		my $line = $line[0];
+		my $len = length($line);
+		for (my $pos=0; $pos < $len; $pos+=80) {
+			print outputfh substr($line, $pos, 80), "\n";
+		}
+	}
+}
+close(outputfh);
+