From 2122bf9ea64c6b305f2d9b5c01f72d04ac243ac0 Mon Sep 17 00:00:00 2001
From: Georgi Tushev <georgi.tushev@brain.mpg.de>
Date: Sun, 4 Nov 2018 22:37:57 +0100
Subject: [PATCH] add cellranger batch

---
 batchCellrangerCounter.pl | 88 +++++++++++++++++++++++++++++++--------
 1 file changed, 70 insertions(+), 18 deletions(-)

diff --git a/batchCellrangerCounter.pl b/batchCellrangerCounter.pl
index ab1d507..de56718 100644
--- a/batchCellrangerCounter.pl
+++ b/batchCellrangerCounter.pl
@@ -8,7 +8,8 @@
 sub version($);
 
 sub createBatchJob($);
-sub runCellranger($$$$);
+sub printSBatchHeader($$$$$$);
+sub printSBatchScript($$$$$);
 
 MAIN:
 {
@@ -19,14 +20,26 @@
     my $path_fastq;
     my $path_genome_index;
     my $path_output;
-    my $opts_cellranger;
+    my $opts;
+    my $sbatch_nodes = 1;
+    my $sbatch_partition = "cuttlefish";
+    my $sbatch_time = "100:00:00";
+    my $sbatch_ntasks = 1;
+    my $sbatch_cpus = 32;
+    my $sbatch_name = "braintest";
     
     # set-up paramters
     Getopt::Long::GetOptions(
         "f|fastq=s" => \$path_fastq,
         "o|output-dir=s" => \$path_output,
         "g|genome=s" => \$path_genome_index,
-        "p|opts=s" => \$opts_cellranger,
+        "x|opts=s" => \$opts,
+        "n|nodes=i" => \$sbatch_nodes,
+        "p|partition=s" => \$sbatch_partition,
+        "t|time=s" => \$sbatch_time,
+        "T|ntasks=i" => \$sbatch_ntasks,
+        "c|cpus=i" => \$sbatch_cpus,
+        "l|name=s" => \$sbatch_name,
         "h|help" => \$help,
         "v|version" => \$version
     ) or usage("Error::invalid command line options");
@@ -41,18 +54,32 @@
     # create batch job
     my $batch_job = createBatchJob($path_fastq);
 
-    # run cellranger count
-    runCellranger($batch_job, $path_output, $path_genome_index, $opts_cellranger);
+    # print SBatch header
+    printSBatchHeader($sbatch_nodes, 
+                      $sbatch_partition, 
+                      $sbatch_time, 
+                      $sbatch_ntasks,
+                      $sbatch_cpus,
+                      $sbatch_name);
+
+    # change working directory
+    print "cd $path_output\n\n";
+
+    # create sbatch script
+    printSBatchScript($batch_job, $path_output, $path_genome_index, $sbatch_cpus, $opts);
     
 }
 
 
-sub runCellranger($$$$)
+sub printSBatchScript($$$$$)
 {
     my $batch_job = $_[0];
     my $path_output = $_[1];
     my $path_genome_index = $_[2];
-    my $opts_cellranger = $_[3];
+    my $sbatch_cpus = $_[3];
+    my $opts = $_[4];
+
+    $path_output =~ s/\/$//g;
 
     foreach my $id (sort keys %{$batch_job})
     {
@@ -61,22 +88,48 @@ ($$$$)
         " --fastqs=" . $batch_job->{$id} .
         " --transcriptome=" . $path_genome_index .
         " --sample=" . $id;
-        $command .= " " . $opts_cellranger if (defined($opts_cellranger));
+        $command .= " " . $opts if (defined($opts));
 
+        # execute cellranger command
+        print "echo $id\n";
         print $command,"\n";
-        system($command);
-
+        
         # move result to output folder
-        if (-d $id)
-        {
-            $path_output =~ s/\/$//g;
-            $command = "mv " . $id  . " " . $path_output . '/' . $id;
-            print $command,"\n";
-            system($command);
-        }
+        $command = "if [ -d  $id]; then\n" .
+        "\tmv $id $path_output/$id\n" .
+        "fi\n";
+        print $command,"\n";
     }
 }
 
+
+sub printSBatchHeader($$$$$$)
+{
+    my $nodes = $_[0];
+    my $partition = $_[1];
+    my $time = $_[2];
+    my $ntasks = $_[3];
+    my $cpus = $_[4];
+    my $name = $_[5];
+
+    # print header
+    print "#!/bin/bash\n";
+    print "\n";
+    print "#SBATCH --nodes=$nodes\n";
+    print "#SBATCH --partition=$partition\n";
+    print "#SBATCH --time=$time\n";
+    print "#SBATCH --ntasks=$ntasks\n";
+    print "#SBATCH --cpus-per-task=$cpus\n";
+    print "#SBATCH --job-name=$name\n";
+    print "#SBATCH --error=error_\%j.out\n";
+    print "#SBATCH --output=output_\%j.out\n";
+    print "\n";
+    print "echo \$SLURM_SUBMIT_DIR\n";
+    print "echo \"Running on \`hostname\`\"\n";
+    print "\n";
+}
+
+
 sub createBatchJob($)
 {
     my $path_fastq = $_[0];
@@ -87,7 +140,6 @@ ($)
     opendir(my $dh, $path_fastq);
     while (my $path_tag = readdir $dh)
     {
-        
         my $path_full = $path_fastq . '/' . $path_tag;
         next unless -d $path_full;
         next if ($path_tag =~ m/^\./);