gatk_split_align

#!/usr/bin/perl
#Programmer: Nicholas Hazekamp
#Date: 6/10/2014

use strict;
use integer;
use Symbol;

my $numargs = $#ARGV + 1;

if ($numargs != 3) {
	print STDERR "Usage: perl gran_split_align <number of reads> <reference file> <sam file>\n";
	exit 1;
}

my $num_reads = shift;
my $ref_file = shift;
my $sam_file = shift;
my $ref = substr $ref_file, 0, -3;
my $sam = substr $sam_file, 0, -4;

my @ref_limits;
my @files;

my $i = 0;
my $position = 0;
my $read_count = 0;
my $num_outputs = 0;

#Open input file
open(INPUT, $ref_file);

open (OUTPUT,">$ref.$num_outputs.fa");
push @ref_limits, 0;

my $file = gensym;
open($file, ">$sam.$num_outputs.sam");
push @files, $file; 

my $last_loc;
while (my $line = <INPUT>) {
	chomp $line;
#FASTQ files begin sequence with '@' character
#If line begins with '@' then it is a new sequence
	if ($line =~ /^>/){
		(my $loc) = $line =~ m/loc(\d+)/;
#Check if the new sequence should be placed in a new file, otherwise place it in same file
		if ($read_count == $num_reads){
			close(OUTPUT);
			$num_outputs++;
			$read_count = 1;

			open(OUTPUT, ">$ref.$num_outputs.fa");
			push @ref_limits, $loc;

			$file = gensym;
			open($file, ">$sam.$num_outputs.sam");
			push @files, $file;

			print OUTPUT "$line\n";
			print "HELP $num_outputs\n";
		}
		else{
			$last_loc = $loc;
			print OUTPUT "$line\n";
			$read_count++;
		}
	}
#place all other lines in FASTQ file under same sequence
	else {
		print OUTPUT "$line\n";
	}
}
push @ref_limits, $last_loc+1;
close(OUTPUT);
close(INPUT);

my $output;

print "START SAM\n";

#Open input file
open(INPUT, $sam_file);

my $in_header = 1;

while (my $line = <INPUT>) {
	chomp $line;

	(my $loc) = $line =~ /loc(\d+)/;
	for( $i = 0; $i <= $num_outputs; $i++ ){
		if($loc >= $ref_limits[$i]){
			$position = $i;
		}
	}
	if($loc eq ""){
		if(not $line =~ /index/){
			foreach $output (@files) {
				print $output "$line\n";
			}
		}
	}
	elsif ($line =~ /^\@SQ/){
		if($in_header eq 0){
			print "\@SQ found after end of header.\n";
			exit;
		}
		$output = $files[$position];
		print $output "$line\n";
	}
	#place all other lines in FASTQ file under same sequence
	else {
		if($in_header == 1){
			$in_header == 0;
		}
		$output = $files[$position];
		print $output "$line\n";
	}
}
close(INPUT);

foreach $output (@files) {
	close($output);
}