RNA-MATE in an SGE queue manager

[inesdesantiago]

Hello
I am trying to use RNA-MATE in a computer cluster with an SGE queue manager. In RNA-MATE v1.1 user guide, the user is instructed to change a couple of things in the RNA-mapping.pm perl script.

But I can't understand what I am supposed to change.
citing the user guide:

line 325 which reads:

print OUT $comm;

should be changed to include the “#$/bin/sh” line, and a newline after the actual command needs
to be inserted.

But what does this mean exactly in terms of changing the script?
Where should one include the “#$/bin/sh” line? Before the 325 line? And what is the right syntax, what should one write in the script to include “#$/bin/sh”?

I am posting also part of the script (from line 296 to 338)

sub mapping{
	my ($argv, $tag_file_name,$map_para, $f_genome) = @_;
       #check the log file before run mapping
        if( $toolobj->check_died() < 0 ){ return -1 }
	my $str = substr($f_genome,rindex($f_genome,'/')+1);
	my ($lib_name,$str1) = split(/\./,$str); 

        # check input file,if input file is zero size, we jump over the mapping process and create an zero size output file.
	my $tag_file = $argv->{'outdir'} . $tag_file_name;
        if( !(-e $tag_file) ){ $toolobj->Log_DIED("file: $tag_file, is not exist! \n"); return -1   }
	my $output = $argv->{outdir}. "$lib_name.$tag_file_name.ma.$map_para->[0].$map_para->[1]";
        if($map_para->[2] == 1) { $output .= ".adj_valid" }
        elsif($map_para->[2] == 2) {$output .= ".adj_all" }

        if( -z $tag_file ){
                $toolobj->Log_WARNING("no tag for genome mapping in $tag_file!\n");
                #create empty files
                open(IN, ">$output");
                close(IN);
                return 1;
        }

        #call mapread program
        my $mymask =  substr($argv->{mask},0,$map_para->[0]);
        my $comm = "$argv->{f2m} -program $argv->{mapreads}  -g  $f_genome ";
        $comm .= "-r $tag_file -d $argv->{outdir} -t $map_para->[0] ";
        $comm .= "-e $map_para->[1] -z $argv->{max_hits} -p $mymask -a $map_para->[2] ";
        my $mysh = "$output.sh";
        open(OUT,">$mysh") or $toolobj->Log_DIED("can't open file $mysh");
        print OUT $comm;
        close(OUT);

        my $success_file = "$output.success";

	#debug
	#$comm = "sh $mysh";
       	$comm = "qsub -l walltime=48:00:00 -o $mysh.out -e $mysh.err $mysh > $mysh.id ";

        my $rc = system($comm);
        my %hash = ($success_file => $rc);

        return %hash;
}

Thank you for your help!
I don't know much about perl, so please remmember this when replying to me...

[dawe]

Hello there,
Never used RNA-MATE but...

        my $comm = "$argv->{f2m} -program $argv->{mapreads}  -g  $f_genome ";
        $comm .= "-r $tag_file -d $argv->{outdir} -t $map_para->[0] ";
        $comm .= "-e $map_para->[1] -z $argv->{max_hits} -p $mymask -a $map_para->[2] ";
        my $mysh = "$output.sh";
        open(OUT,">$mysh") or $toolobj->Log_DIED("can't open file $mysh");
        print OUT $comm;
        close(OUT);

        my $success_file = "$output.success";

	#debug
	#$comm = "sh $mysh";
       	$comm = "qsub -l walltime=48:00:00 -o $mysh.out -e $mysh.err $mysh > $mysh.id ";

        my $rc = system($comm);

The perl code builds a bash script to be queued in SGE. the variable $comm contains all the instructions to run RNA-MATE. You may either add this line before 325:

print OUT "#$/bin/sh\n\n";

or modify the $comm construction to have "#$/bin/sh" at the beginning of the string.
Also, I don't know which SGE version you are using, but AFAIK the correct syntax to specify SGE shell is not the one suggested. I believe the correct patch should be:

print OUT "#$ -S /bin/sh\n\n";

For the same reason you may also make the whole thing work not adding that line at all but simply modifying the "exec" line below like this:

       	$comm = "qsub -S /bin/sh -l walltime=48:00:00 -o $mysh.out -e $mysh.err $mysh > $mysh.id ";

which is much more clear to me :-)


HTH
d

[inesdesantiago]

Hi dawe!
Thanks for the reply! I tried with your last suggestions

$comm = "qsub -S /bin/sh -l walltime=48:00:00 -o $mysh.out -e $mysh.err $mysh > $mysh.id ";

and it worked perfectly. I would never be able to think about the -S thing by myself, so thank you!

Just a small note for whoever reads this post:
for SGE systems the "walltime=48:00:00" has to be changed as well (this is refered in the user guide and has to be changed to "s_rt=48:00:00"), i.e, the full line I used and that worked was:

$comm = "qsub -S /bin/sh -l s_rt=48:00:00 -o $mysh.out -e $mysh.err $mysh > $mysh.id ";

Thanks!
Ines