psortb

#!/usr/bin/perl
# Perl script wrapper to run a "docker run" command which runs 
# PSORTb inside the docker container.

use strict;
use Getopt::Long;
use File::Basename;
use Cwd 'abs_path';

our $docker_results_dir = "/tmp/results";
our $docker_image = "brinkmanlab/psortb_commandline:1.0.2";

sub main {

    my ($seqfile, $help, $nondocker_results_dir);
    my ($positive, $negative, $archaea, $xskiploc);
    my ($format, $verbose, $cutoff, $exact, $version, $output, $divergent);

    my @args = @ARGV;
    my $extra_args = remove_unneeded_args(\@args);


    GetOptions('seq|i=s'     => \$seqfile,  'positive|p'    => \$positive,
               'archaea|a'   => \$archaea,  'negative|n'    => \$negative,
               'x-skip-localization' => \$xskiploc,
               'help|h'      => \$help,     'outdir|r=s'    => \$nondocker_results_dir,
               'format|f=s'  => \$format,   'divergent|d=f' => \$divergent,
               'verbose|v'   => \$verbose, 
               'cutoff|c=f'  => \$cutoff,   'output|o=s'    => \$output,
               'exact|e'     => \$exact,    'version'       => \$version);

    usage() if($help);

    if ($version) {
        my $cmd = "sudo docker run --rm -ti $docker_image /usr/local/psortb/bin/psort --version";
        #print "cmd = $cmd\n";
        system($cmd);
        exit(0);
    }

    if (! -d $nondocker_results_dir) {
        print "Please enter a valid local directory to store the PSORTb results in (-r).\n";
        usage();
    }
    elsif (! -e $seqfile) {
        print "Please enter a valid FASTA-format protein sequence file (-i).\n";
    }

    $seqfile = abs_path($seqfile);
    $nondocker_results_dir  = abs_path($nondocker_results_dir); 
    $nondocker_results_dir =~ s/\/$//; # remove trailing slash if present

    my $seqfilename = basename($seqfile);
    my $nondocker_seqfile = "$nondocker_results_dir/$seqfilename";
    my $docker_seqfile    = "$docker_results_dir/$seqfilename";

    # copy file to mount directory (if it doesn't already exist)
    my $seqfile_copied = make_file_copy($seqfile, $nondocker_seqfile);
    

    my $cmd = "sudo docker run --rm -v $nondocker_results_dir:$docker_results_dir -e MOUNT='$nondocker_results_dir' -ti $docker_image /usr/local/psortb/bin/psort $extra_args -i $docker_seqfile";
    #print "cmd = $cmd\n";
    system($cmd);

    # remove copied files (if they were created)
    delete_file($nondocker_seqfile, $seqfile_copied);
}

sub make_file_copy {
    my ($orig_file, $new_file) = @_;

    my $cmd = "cp $orig_file $new_file";
    system($cmd);
    die "Could not copy file: $cmd: $@\n" if $@;
}

sub delete_file {
    my ($file, $delete_flag) = @_;

    if ($delete_flag) {
        my $cmd = "rm $file";
        system($cmd);
        die "Could not delete file: $cmd: $@\n" if $@;
    }
}

sub remove_unneeded_args {
    my $args = shift;

    # remove -r (non-docker results dir) because inside docker, the 
    # "Saving results to <-r option value>" is provided by an environment
    # variable.
    # Remove -i <local file> (because we need to use -i <container file>)
    my @modified_arg_list;

    for (my $i=0; $i<scalar(@$args); $i++) {

        if (($$args[$i] eq '-r') ||
            ($$args[$i] eq '--outdir') ||
            ($$args[$i] eq '-i') || 
            ($$args[$i] eq '--seq')) {
            $i++;
        }
        else {
            push(@modified_arg_list, $$args[$i]);
        }
    }
    return( join(" ", @modified_arg_list) );
}

sub usage {
    $0 =~ /^.*\/(.+)/;
    print("\nUsage: $1 -i <sequence file> -r <local results directory> [-n|-p|-a] [OPTIONS]\n");
    print("Example command: psortb -n -i myseqs.fasta -r /tmp/psortb_results\n");
    print("  --seq, -i         Input sequence file path (required)\n");
    print("  --positive, -p    Gram positive bacteria\n");
    print("  --negative, -n    Gram negative bacteria\n");
    print("  --archaea, -a     Archaea\n");
    print("  --outdir, -r      Path of where to save results files.\n");
    print("  --cutoff, -c      Sets a cutoff value for reported results\n");
    print("  --divergent, -d   Sets a cutoff value for the multiple\n");
    print("                    localization flag\n");
    print("  --format, -f      Specifies sequence format (default is FASTA)\n");
    print("  --exact, -e       Skip SCLBLASTe (useful for batch runs of data\n");
    print("                    against itself in SCLBLAST)\n");
    print("  --output, -o      Specifies the format for the output (default is\n");
    print("                    'normal'). Value can be one of: normal, terse or long\n");
    print("  --verbose, -v     Be verbose while running\n");
    print("  --version         Print the version of PSortb\n");
    print("  --x-skip-localization  Comma-separated list of localizations to skip\n");
    print("  --help, -h        Displays usage information\n\n");

    exit(0);
}

main();