152 lines
3.5 KiB
Perl
Executable File
152 lines
3.5 KiB
Perl
Executable File
#!/usr/bin/perl -w
|
|
|
|
# Runs a given GATK walker genome-wide, but splits up the jobs
|
|
# and then merges the results together. One should really use
|
|
# scatter-gather here, but I wanted to test against that framework
|
|
# (to ensure consistency) so I wrote this temporary script.
|
|
# Also, it allowed me to add in farm commands for wait ids
|
|
# (which is currently unavailable in scatter-gather).
|
|
# If intervals file is left blank, it splits by chromosome.
|
|
|
|
use strict;
|
|
use Getopt::Long;
|
|
|
|
sub usage {
|
|
print "Usage: perl enqueueGATKjobsByChromosome.pl\n\t-cmd <GATK command without output or -L args>\n\t-o <output file>\n\t[-oarg GATK output argument; default:o]\n\t[-i intervals file]\n\t[-n number of splits to make]\n\t[-q farm queue; default:gsa]\n\t[-wait farm wait id]\n\t[-job farm job name]\n\t[-dry]\n\t[-bam is output format bam?; default:no]\n";
|
|
exit(1);
|
|
}
|
|
|
|
my $cmd = undef;
|
|
my $output = undef;
|
|
my $outputArg = "o";
|
|
my $jobName = undef;
|
|
my $wait = undef;
|
|
my $queue = "gsa";
|
|
my $breaks = 24;
|
|
my $intervalsFile = undef;
|
|
my $dry;
|
|
my $isBam;
|
|
|
|
GetOptions( "cmd=s" => \$cmd,
|
|
"o=s" => \$output,
|
|
"oarg:s" => \$outputArg,
|
|
"job:s" => \$jobName,
|
|
"wait:s" => \$wait,
|
|
"n:i" => \$breaks,
|
|
"i:s" => \$intervalsFile,
|
|
"dry!" => \$dry,
|
|
"bam!" => \$isBam,
|
|
"q:s" => \$queue);
|
|
|
|
usage() if ( !$cmd || !$output );
|
|
|
|
my @intervals;
|
|
if (!$intervalsFile) {
|
|
my $num = 1;
|
|
while ($num < 23) {
|
|
push(@intervals, $num);
|
|
$num++;
|
|
}
|
|
push(@intervals, "X");
|
|
push(@intervals, "Y");
|
|
} else {
|
|
open(FILE, "< $intervalsFile") or die "can't open $intervalsFile: $!";
|
|
my @lines = <FILE>;
|
|
chomp(@lines);
|
|
my $linecount = scalar(@lines);
|
|
if ($linecount < $breaks) {
|
|
$breaks = $linecount;
|
|
}
|
|
|
|
my $linesPerJob = $linecount / $breaks;
|
|
my $index = 0;
|
|
for (my $i = 1; $i < $breaks; $i++) {
|
|
my $interval = "";
|
|
for (my $j = 0; $j < $linesPerJob; $j++) {
|
|
$interval .= "$lines[$index];";
|
|
$index++;
|
|
}
|
|
push(@intervals, $interval);
|
|
}
|
|
my $interval = "";
|
|
while ($index < $linecount) {
|
|
$interval .= "$lines[$index];";
|
|
$index++;
|
|
}
|
|
push(@intervals, $interval);
|
|
}
|
|
|
|
my $intervalcount = scalar(@intervals);
|
|
for (my $i = 0; $i < $intervalcount; $i++) {
|
|
enqueue($intervals[$i], $cmd, $output, $outputArg, $wait, $queue, $dry, $isBam, ($i+1));
|
|
}
|
|
|
|
mergeResults($output, $queue, $dry, $isBam, $jobName, $intervalcount);
|
|
|
|
sub enqueue {
|
|
|
|
my $interval = $_[0];
|
|
my $cmd = $_[1];
|
|
my $outArg = $_[3];
|
|
my $waitid = $_[4];
|
|
my $queue = $_[5];
|
|
my $dry = $_[6];
|
|
my $bam = $_[7];
|
|
my $index = $_[8];
|
|
|
|
my $output = "$_[2].$index";
|
|
if ($bam) {
|
|
$output .= ".bam";
|
|
}
|
|
|
|
my $bsub = "bsub -q $queue -o $output.sdout -J $_[2]";
|
|
if ($waitid) {
|
|
$bsub .= " -w \"ended($waitid)\"";
|
|
}
|
|
|
|
my $command = "$bsub $cmd -$outputArg $output -L $interval";
|
|
execute($command, $dry);
|
|
}
|
|
|
|
sub mergeResults {
|
|
|
|
my $output = $_[0];
|
|
my $queue = $_[1];
|
|
my $dry = $_[2];
|
|
my $bam = $_[3];
|
|
my $jobName = $_[4];
|
|
my $intervalcount = $_[5];
|
|
|
|
my $cmd = "bsub -q $queue -o $output.sdout -w \"ended($output)\"";
|
|
if ($jobName) {
|
|
$cmd .= " -J $jobName";
|
|
}
|
|
|
|
if ($bam) {
|
|
$cmd .= " samtools merge $output ";
|
|
for (my $i = 1; $i <= $intervalcount; $i++) {
|
|
$cmd .= "$output.$i.bam ";
|
|
}
|
|
} else {
|
|
$cmd .= " cat ";
|
|
for (my $i = 1; $i <= $intervalcount; $i++) {
|
|
$cmd .= "$output.$i ";
|
|
}
|
|
$cmd .= "> $output";
|
|
}
|
|
|
|
execute($cmd, $dry);
|
|
}
|
|
|
|
sub execute {
|
|
|
|
my $cmd = $_[0];
|
|
my $dry = $_[1];
|
|
|
|
if ($dry) {
|
|
print "$cmd\n";
|
|
} else {
|
|
system($cmd);
|
|
}
|
|
}
|