From da87c9a28dd1e8fbc564c2dfad0100f4218c1411 Mon Sep 17 00:00:00 2001
From: elserj <elserj@localhost>
Date: Thu, 3 Dec 2015 01:20:45 +0000
Subject: [PATCH] Fixed all bugs and also now outputs the lines to a set of
 files.  Probably should find a better way to name the output files, but can
 be changed later.

svn path=/; revision=641
---
 interactome_scripts/SNP_venn_stats.pl | 67 ++++++++++++++++++++++-----
 1 file changed, 56 insertions(+), 11 deletions(-)

diff --git a/interactome_scripts/SNP_venn_stats.pl b/interactome_scripts/SNP_venn_stats.pl
index 9ef9d98..cc8a97b 100755
--- a/interactome_scripts/SNP_venn_stats.pl
+++ b/interactome_scripts/SNP_venn_stats.pl
@@ -18,8 +18,8 @@ my %file1hash;
 my %file2hash;
 my %file3hash;
 
-# This will change the way the hash is keyed.  A "0" will key with only the position, a "1" will include the snp value
-my $include_snp = 0;
+# This will change the way the hash is keyed.  A "0" will key with only the position, a "1" will include the reference and snp value
+my $include_snp = 1;
 
 open(FILE1, "$file_1");
 while(<FILE1>) {
@@ -28,10 +28,10 @@ while(<FILE1>) {
 		my ($chrom,$pos,$ref,$snp) = split("\t", $line);
 		my $key;
 		
-		if($include_snp) {
-				$key = "$chrom" . "x" . "$pos";
+		if($include_snp == 0) {
+				$key = "$chrom" . "xxx" . "$pos";
 		}else{
-				$key = "$chrom" . "x" . "$pos" . "y" . "$snp";
+				$key = "$chrom" . "xxx" . "$pos" . "xxx" . "$ref". "xxx" . "$snp";
 		}
 		$file1hash{$key} = $snp;
 }
@@ -45,10 +45,10 @@ while(<FILE2>) {
 		my ($chrom,$pos,$ref,$snp) = split("\t", $line);
 		my $key;
 		
-		if($include_snp) {
-				$key = "$chrom" . "x" . "$pos";
+		if($include_snp == 0) {
+				$key = "$chrom" . "xxx" . "$pos";
 		}else{
-				$key = "$chrom" . "x" . "$pos" . "y" . "$snp";
+				$key = "$chrom" . "xxx" . "$pos" . "xxx" . "$ref". "xxx" . "$snp";
 		}
 		$file2hash{$key} = $snp;
 }
@@ -61,10 +61,10 @@ while(<FILE3>) {
 		my ($chrom,$pos,$ref,$snp) = split("\t", $line);
 		my $key;
 		
-		if($include_snp) {
-				$key = "$chrom" . "x" . "$pos";
+		if($include_snp == 0) {
+				$key = "$chrom" . "xxx" . "$pos";
 		}else{
-				$key = "$chrom" . "x" . "$pos" . "y" . "$snp";
+				$key = "$chrom" . "xxx" . "$pos" . "xxx" . "$ref". "xxx" . "$snp";
 		}
 		$file3hash{$key} = $snp;
 }
@@ -80,20 +80,38 @@ my $count_file_23 = 0;
 
 my $count_file_123 = 0;
 
+# add the output files to put out each line
+open(OUT1, ">file_1_only.txt");
+open(OUT2, ">file_2_only.txt");
+open(OUT3, ">file_3_only.txt");
+open(OUT12, ">file_1_and_2.txt");
+open(OUT13, ">file_1_and_3.txt");
+open(OUT23, ">file_2_and_3.txt");
+open(OUT123, ">file_1__and_2_and_3.txt");
+
 foreach my $key (keys %file1hash) {
 		#check if also in file2
 		if(defined($file2hash{$key})) {
 				# and file3
 				if(defined($file3hash{$key})) {
 						$count_file_123++;
+						my $output = split_key($key);
+						print OUT123 "$output\n";
+						
 				}else{# it's only in file1 and file2
 						$count_file_12++;
+						my $output = split_key($key);
+						print OUT12 "$output\n";
 				}
 				#check if in file3
 		}elsif(defined($file3hash{$key})) {
 				$count_file_13++;
+				my $output = split_key($key);
+					print OUT13 "$output\n";
 		#it's only in file1
 		}else{$count_file_1++;
+				my $output = split_key($key);
+				print OUT1 "$output\n";
 		}
 }
 
@@ -101,9 +119,28 @@ foreach my $key (keys %file2hash) {
 		#check if in file3
 		if(defined($file3hash{$key}) && !defined($file1hash{$key})) {
 				$count_file_23++;
+				my $output = split_key($key);
+				print OUT23 "$output\n";
+		}elsif(!defined($file3hash{$key}) && !defined($file1hash{$key})) {
+				my $output = split_key($key);
+				print OUT2 "$output\n";
+		}
+}
+
+foreach my $key (keys %file3hash) {
+		if(!defined($file1hash{$key}) && !defined($file2hash{$key})) {
+				my $output = split_key($key);
+				print OUT3 "$output\n";
 		}
 }
 
+close(OUT1);
+close(OUT2);
+close(OUT3);
+close(OUT12);
+close(OUT13);
+close(OUT23);
+close(OUT123);
 
 # Get total counts in each file/hash
 my $total_count_file_1 = scalar(keys %file1hash);
@@ -129,3 +166,11 @@ print OUTFILE "In $file_2 and $file_3\t=\t$count_file_23\n";
 print OUTFILE "In all three files\t=\t$count_file_123\n";
 
 close(OUTFILE);
+
+sub split_key{
+		my $key = $_[0];
+		my ($chrom, $pos, $ref, $snp) = split("xxx", $key);
+		my $string = "$chrom\t$pos\t$ref\t$snp";
+		return $string;
+}
+		
-- 
2.34.1