From 761fad0ab5f5c68f3b462d72ba8fc9676fdf0354 Mon Sep 17 00:00:00 2001
From: preecej <preecej@localhost>
Date: Wed, 23 May 2012 20:43:06 +0000
Subject: [PATCH] Simple gene and expression data mapping scripts for the Rice
 Seed Dev project

svn path=/; revision=334
---
 .../preecej/perl_singletons/affy_mapping.pl   | 67 +++++++++++++++++++
 .../preecej/perl_singletons/ath_mapping.pl    | 59 ++++++++++++++++
 2 files changed, 126 insertions(+)
 create mode 100644 Personnel/preecej/perl_singletons/affy_mapping.pl
 create mode 100644 Personnel/preecej/perl_singletons/ath_mapping.pl

diff --git a/Personnel/preecej/perl_singletons/affy_mapping.pl b/Personnel/preecej/perl_singletons/affy_mapping.pl
new file mode 100644
index 0000000..da2423e
--- /dev/null
+++ b/Personnel/preecej/perl_singletons/affy_mapping.pl
@@ -0,0 +1,67 @@
+#!/usr/bin/perl -w
+# maps affy chip expression data to gene loci, writes to file, and sorts
+
+use strict;
+use Data::Dumper;
+
+my %gene_list;
+
+open GENES,$ARGV[0];
+readline GENES;
+
+while (<GENES>) {
+	chomp;
+	my @curr_ary = split('\t',$_);
+
+	$curr_ary[0] =~ s/\s+$//; # rtrim whitespace
+	$curr_ary[0] =~ s/^\s+//; # ltrim whitespace
+	$curr_ary[1] =~ s/\s+$//; # rtrim whitespace
+	$curr_ary[1] =~ s/^\s+//; # ltrim whitespace
+
+	$gene_list{$curr_ary[1]}{gene_id} = $curr_ary[0];
+	
+	#print "$curr_ary[0]\n";
+}
+
+close GENES;
+
+#print Dumper(\%gene_list) . "\n\n";
+
+open EXPR, $ARGV[1];
+readline EXPR;
+
+while (<EXPR>) {
+	chomp;
+	my @expr_ary = split('\t',$_);
+	if (exists $gene_list{$expr_ary[0]}) {
+		$gene_list{$expr_ary[0]}{value} = $expr_ary[1];
+		$gene_list{$expr_ary[0]}{abs_call} = $expr_ary[2];
+		$gene_list{$expr_ary[0]}{p_value} = $expr_ary[3];
+	}
+	
+	#print "$expr_ary[0]\n";
+}
+
+close EXPR;
+
+#print Dumper(\%gene_list);
+
+open OUTFILE, ">mapped_$ARGV[1]";
+
+foreach my $mapping (keys %gene_list) {
+	if (exists $gene_list{$mapping}{value}) {
+		print OUTFILE "$gene_list{$mapping}{gene_id}\t"
+			. "$mapping\t"
+			. "$gene_list{$mapping}{value}\t"
+			. "$gene_list{$mapping}{abs_call}\t"
+			. "$gene_list{$mapping}{p_value}\n";
+	} else {
+		print OUTFILE "$gene_list{$mapping}{gene_id}\t$mapping\tnone\tnone\tnone\n"
+	}
+}
+
+close OUTFILE;
+
+system "cat ./mapped_$ARGV[1] | sort | uniq > ../mapping_output/sorted_mapped_$ARGV[1]; rm ./mapped_$ARGV[1]";
+
+exit(0);
diff --git a/Personnel/preecej/perl_singletons/ath_mapping.pl b/Personnel/preecej/perl_singletons/ath_mapping.pl
new file mode 100644
index 0000000..b68a284
--- /dev/null
+++ b/Personnel/preecej/perl_singletons/ath_mapping.pl
@@ -0,0 +1,59 @@
+#!/usr/bin/perl -w
+# maps A.th. expression data to gene loci, writes to file, and sorts
+use strict;
+use Data::Dumper;
+
+my %gene_list;
+
+open GENES,$ARGV[0];
+readline GENES;
+
+while (<GENES>) {
+	chomp;
+
+	$_ =~ s/\s+$//; # rtrim whitespace
+        $_ =~ s/^\s+//; # ltrim whitespace
+
+	$gene_list{$_} = "";
+}
+
+close GENES;
+
+#print Dumper(\%gene_list) . "\n\n";
+
+open EXPR, $ARGV[1];
+readline EXPR;
+
+while (<EXPR>) {
+	#print uc($_);
+	chomp;
+	my @expr_ary = split('\t',$_);
+
+	$expr_ary[0] =~ s/\s+$//; # rtrim whitespace
+        $expr_ary[0] =~ s/^\s+//; # ltrim whitespace
+	$expr_ary[1] =~ s/\s+$//; # rtrim whitespace
+        $expr_ary[1] =~ s/^\s+//; # ltrim whitespace
+
+	#print $expr_ary[0] . "|" . $expr_ary[1] . "\n";
+	if (exists $gene_list{uc($expr_ary[0])}) {
+		$gene_list{uc($expr_ary[0])} = $expr_ary[1];
+	}
+}
+
+close EXPR;
+
+#print Dumper(\%gene_list);
+
+open OUTFILE, ">mapped_$ARGV[1]";
+
+foreach my $mapping (keys %gene_list) {
+	if ($gene_list{$mapping} ne "") {
+		print OUTFILE "$mapping\t$gene_list{$mapping}\n"
+	}
+}
+
+close OUTFILE;
+
+system "cat ./mapped_$ARGV[1] | sort | uniq > ./mapping_output/sorted_mapped_$ARGV[1]; rm ./mapped_$ARGV[1]";
+
+exit(0);
-- 
2.34.1