From b8018903e38a18b6aa6cc7ab3e4924c192d79352 Mon Sep 17 00:00:00 2001
From: athreyab <athreyab@localhost>
Date: Tue, 1 May 2012 00:20:22 +0000
Subject: [PATCH] bash script for creating singleton/pair

svn path=/; revision=326
---
 .../athreyab/illumina_parse/create_pairs      | 117 ++++++++++++++++++
 .../athreyab/illumina_parse/float_compare.pl  |  12 ++
 2 files changed, 129 insertions(+)
 create mode 100755 Personnel/athreyab/illumina_parse/create_pairs
 create mode 100644 Personnel/athreyab/illumina_parse/float_compare.pl

diff --git a/Personnel/athreyab/illumina_parse/create_pairs b/Personnel/athreyab/illumina_parse/create_pairs
new file mode 100755
index 0000000..8626f03
--- /dev/null
+++ b/Personnel/athreyab/illumina_parse/create_pairs
@@ -0,0 +1,117 @@
+#!/bin/bash
+
+function process_file{
+	array=("$@")
+	for i in "${array[@]}"
+	do	
+		echo "processing file ${array[i]}"
+		while read line ; do
+			if [ $line_index == 0 ]
+			then
+				if [[ "$line" == *1:Y:* ]]
+				then
+					out=""
+				else
+					out="$out$line"
+				fi
+				line_index=$((line_index+1))
+			elif [ $line_index == 1 ]
+			then
+				if [ -z "$out" ]
+				then
+					out=""
+				else
+					ns="${line//[^C]}"
+					perl float_compare.pl "${#line}" "${#ns}" "${max_rate}"
+					r=$?
+					if [ $r -eq 1 ]
+					then								
+						out="$out||$line"
+						echo $out >> "$output_folder/file_$i";
+						out=""
+					else
+						out=""
+					fi
+				fi
+				line_index=$((line_index+1))
+			elif [ $line_index == 2 ]
+			then		
+				line_index=$((line_index+1))
+
+			elif [ $line_index == 3 ]
+			then		
+				line_index=0
+			fi
+		done < "${array[i]}"
+	done
+}
+
+input_folder=$1
+output_folder=$2
+max_rate=$3
+
+echo -e "max_rate is $max_rate \n"
+
+#check if the input and output folder are defined in command line
+if [ -z "$input_folder" ]
+then
+	echo -e "Input folder is not defined \n"
+	exit;	
+elif [ -z "$output_folder" ]
+then
+	echo -e "Output folder is not defined \n"
+	exit;
+elif [ -z "$max_rate" ]
+then
+	echo -e "max rate is not defined \n"
+
+else
+	if [ -d "$output_folder" ]; then
+	    	echo "output directory already exists. Exiting now!"
+		exit;
+	fi
+
+	mkdir $output_folder
+	#create two temp files - array1.txt and array2.txt. array1.txt contains all files from reading 1 in sorted order
+	#array2.txt  contains all files from reading 2 in sorted order
+	ls $input_folder*R1*.fastq | sort > array1.txt
+	ls $input_folder*R2*.fastq | sort > array2.txt
+	
+	#put the fileNames from reading 1 in array - array1
+	index1=0
+	while read line ; do
+		array1[$index1]="$line"
+		index1=$(($index1+1))
+	done < array1.txt
+
+	#put the fileNames from reading 2 in array - array2
+	index2=0
+        while read line ; do
+                array2[$index2]="$line"
+                index2=$(($index2+1))
+	done < array2.txt
+
+	#  if array1 and array2 are not of equal size, then
+	# some files are missing.
+	if [ $index1 != $index2 ]
+	then
+		echo "some readings are missing\n";
+		exit;
+	fi
+
+	# delete temp files as we dont' need them anymore. We already
+	# have that information in array1 and array2
+	rm -f array1.txt
+	rm -f array2.txt
+
+	i=0
+	line_index=0
+	out=""
+	remove_entry=0
+	
+	# loop through each files in array1, filter them and put in a file
+	process_file "${array1[@]}"
+	process_file "${array2[@]}"
+fi
+
+
diff --git a/Personnel/athreyab/illumina_parse/float_compare.pl b/Personnel/athreyab/illumina_parse/float_compare.pl
new file mode 100644
index 0000000..603f89f
--- /dev/null
+++ b/Personnel/athreyab/illumina_parse/float_compare.pl
@@ -0,0 +1,12 @@
+#!/usr/bin/perl
+
+$n1 = $ARGV[0];
+$n2 = $ARGV[1];
+$n3 = $ARGV[2];
+
+if($n2/$n1 <= $n3){
+	exit 1;
+}
+else{
+	exit 0;
+}
-- 
2.34.1