From e2bdb1c70b4d33990dd17fe4250e3c3424a2a00d Mon Sep 17 00:00:00 2001
From: athreyab <athreyab@localhost>
Date: Mon, 16 Apr 2012 20:47:29 +0000
Subject: [PATCH] illumina parse scripts

svn path=/; revision=325
---
 .../athreyab/illumina_parse/illumina_parse.pl | 121 ++++++++++++++++++
 .../interactions/interactionPathsFromTsv.pl   |   4 +-
 2 files changed, 123 insertions(+), 2 deletions(-)
 create mode 100755 Personnel/athreyab/illumina_parse/illumina_parse.pl

diff --git a/Personnel/athreyab/illumina_parse/illumina_parse.pl b/Personnel/athreyab/illumina_parse/illumina_parse.pl
new file mode 100755
index 0000000..d876a78
--- /dev/null
+++ b/Personnel/athreyab/illumina_parse/illumina_parse.pl
@@ -0,0 +1,121 @@
+#!/usr/bin/perl
+
+use Switch;
+
+
+$optionId = $ARGV[0];
+
+if($optionId eq "-co" && ($#ARGV == 2 || $#ARGV == 3)){
+	$in_folder = $ARGV[1];
+	$out_folder = $ARGV[2];
+	$slash_string = $ARGV[3];
+	if($slash_string eq ""){
+		$slash_string = '\\';	
+	}
+	process_folders($in_folder,$out_folder,"",$slash_string);
+}
+
+elsif($optionId eq "-fi" && ($#ARGV == 3 || $#ARGV == 4)){
+	$in_folder = $ARGV[1];
+	$out_folder = $ARGV[2];
+	$max_rate = $ARGV[3];
+	$slash_string = $ARGV[4];
+	if($max_rate =~ /[^0-9\.]/){
+		print "\nmax_rate supplied: '$max_rate', is not a number. check if you supplied all 4 params - option(-fi), an input directory, output directory and a max_rate \n";
+		showUsage();
+	}
+	if($slash_string eq ""){
+		$slash_string = '\\';	
+	}
+	process_folders($in_folder,$out_folder,$max_rate,$slash_string);
+}
+
+else{
+	print "Incorrect parameters were supplied\n";
+	showUsage();
+}
+
+sub showUsage
+{
+	print "\nUsage: \n";
+	print "convert fastq to fasta format:\t./illumina_parse -co <input_path> <output_path> [<slash_string>]\n";
+	print "filter low quality entries:\t./illumina_parse -fi <input_path> <output_path> <max_rate> [<slash_string>]\n";
+	print "[] = optional parameter\n";
+	die "";	
+}
+
+sub isnan { ! defined( $_[0] <=> 9**9**9 ) }
+
+sub filterRead {
+        my $seq         =shift;
+        my $maxRate     =shift;
+        my $length      =length($seq);
+        my $count       =0;
+        $count++ while $seq=~m/N/g;
+	$count++ while $seq=~m/\./g;
+        return 1 if(($count/$length)<=$maxRate);
+        return 0;
+}
+
+sub process_folders{
+	$in_folder = $_[0];
+	$out_folder = $_[1];
+	$max_rate = $_[2];
+	$slash_string = $_[3];
+	opendir(DIR, $in_folder) or die "can't open directory $in_folder: $!\n";
+	unless(-d $out_folder){
+ 	   mkdir $out_folder or die "can't create directory $out_folder: $! \n";
+	}
+	@files = readdir DIR;
+	$y_entries = 0;
+	$total_entries = 0;
+	$high_qual_entries = 0;
+	foreach $fileName (@files) {
+		if(index($fileName,'.fastq') != -1){
+	 		open(in_file, "<","$in_folder/$fileName") || die "Error: file '$in_folder/$fileName' can not be opened\n";
+			print "processing file $in_folder/$fileName\n";
+			$fileName =~ s/.fastq/.fa/;
+			open(out_file,">>","$out_folder/$fileName") || die "Error: file $out_folder/$fileName can not be opened\n";
+			until(eof(in_file)){
+	 	      		$h1=<in_file>;
+				$seq=<in_file>;
+		        	$h2=<in_file>;
+        			$qual=<in_file>;
+				$total_entries = $total_entries + 1;
+        			chomp $h1;
+	        		chomp $seq;
+        			$h1=~s/\@/>/;
+				if(index($h1, ' 1:Y') != -1 || index($h1,' 2:Y') != -1){
+                                        $y_entries = $y_entries + 1;
+                                        next;
+				}
+				if(index($h1, ' 1:N') != -1){
+					$temp = $slash_string."1:N";
+					$h1 =~ s/ 1:N/$temp/;
+					$h1 = $h1.$slash_string."1";
+				}
+				elsif(index($h1, ' 2:N') != -1){
+					$temp = $slash_string."2:N";
+					$h1 =~ s/ 2:N/$temp/;
+					$h1 = $h1.$slash_string."2";
+				}
+							
+				if($max_rate eq ""){
+					print out_file "$h1\n$seq\n";
+				}
+				elsif(filterRead($seq,$max_rate)){
+					print out_file "$h1\n$seq\n";
+					$high_qual_entries = $high_qual_entries + 1;
+				}
+			}
+		close in_file;
+		close out_file;
+		print "successfully written $fileName\n";
+ 		}
+	} 	
+	print "total entries: $total_entries\n";
+	print "low quality entries(entries containing 1:Y or 2:Y in the header): $y_entries\n";
+	$qual_entries = $total_entries - $high_qual_entries;
+	print "entries removed due to exceeding max_rate: $qual_entries \n";
+	closedir DIR;
+}
diff --git a/Personnel/athreyab/interactions/interactionPathsFromTsv.pl b/Personnel/athreyab/interactions/interactionPathsFromTsv.pl
index 950f0aa..c81d3dd 100644
--- a/Personnel/athreyab/interactions/interactionPathsFromTsv.pl
+++ b/Personnel/athreyab/interactions/interactionPathsFromTsv.pl
@@ -75,11 +75,11 @@ sub importSeedInteractionData(){
      	print "interaction_id:$interaction_id\n";
      	#if not, insert data into to the database
      	if(isEmpty($interaction_id)){
-      	$query = "INSERT INTO Interaction(`object_id_left`,`object_id_right`,`interaction_type_id`,`mode_of_action_id`,
+      		$query = "INSERT INTO Interaction(`object_id_left`,`object_id_right`,`interaction_type_id`,`mode_of_action_id`,
       	`interactor_type_id_left`,`interactor_type_id_right`,`evidence_id`,`evidence_code_id`,`experiment_id`,`comments`,`curator_id`,`date`) 
       	VALUES ('$obj_id_left','$obj_id_right','$interaction_type_id','$mode_of_action_id','$int_type_id_left','$int_type_id_right'
       	,'$evidence_id','$evidence_code_id','$experiment_id','$comments','$curator_id', '".strftime("%Y-%m-%d", localtime)."')";      
-      	executeDbQuery($query);
+      		executeDbQuery($query);
       }				
 	}
 }
-- 
2.34.1