psortb_app

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

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

our $container_results_dir = "/results";
our $sif_file = "./psortb.sif";
our $overlay_size = 300; # MB
our $overlay_file = "results.img";

sub main {

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

    my @args = @ARGV;

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

    my $ret = GetOptions('seq|i=s'       => \$seqfile,
               'positive|p'    => \$positive,
               'archaea|a'     => \$archaea,
               'negative|n'    => \$negative,
               'x-skip-localization' => \$xskiploc,
               'divergent|d=f' => \$divergent,
               'format|f=s'    => \$input_format,
               'cutoff|c=f'    => \$cutoff,
               'exact|e'       => \$exact,
               'outdir|r=s'        => \$host_results_dir,
	             'sif-file|s=s'      => \$sif_file,
               'overlay-size-mb=i' => \$overlay_size,
               'overlay-file=s'    => \$overlay_file,
               'output|o=s'        => \$output,
               'version'           => \$version,
               'help|h' => \$help, 'verbose|v' => \$verbose);

    usage() if($help);
    die "Exiting" unless $ret; # die if unrecognized option is detected

    my $extra_args = remove_unneeded_args(\@args);

    if ($version) {
        my $cmd = "apptainer run $sif_file /usr/local/psortb/bin/psort --version";
        #print "cmd = $cmd\n";
        system($cmd);
        exit(0);
    }

    if (! -e $sif_file) {
        print "ERROR: Apptainer image file (sif file) not found: $sif_file\nPlease enter an apptainer sif file location in (-s).\n";
        exit(0);
    }
    elsif (! -d $host_results_dir) {
        print "ERROR: Directory not found: $host_results_dir\nPlease enter a valid local directory to store the PSORTb results in (-r).\n";
        exit(0);
    }
    elsif (! -e $seqfile) {
        print "ERROR: Sequence file not found: $seqfile\nPlease enter a valid FASTA-format protein sequence file (-i).\n";
        exit(0);
    }

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

    my $seqfilename       = basename($seqfile);
    my $host_seqfile      = "$host_results_dir/$seqfilename";
    my $container_seqfile = "$container_results_dir/$seqfilename";

    my ($hname, $hpath, $hsuffix) = fileparse($host_results_dir);
    my @hdirs = split(/\//, $hpath);
    my $host_results_rootdir = "/$hdirs[1]";

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

    # create overlay image to save results to initially
    my $overlay_created_flag = 0;
    if (! -e $overlay_file) {
      my $cmd1 = "apptainer overlay create --create-dir $container_results_dir --size $overlay_size $overlay_file 2> psortb.err";
      if ($verbose) {
        print "Running: $cmd1\n\n";
      }
      system($cmd1);
      $overlay_created_flag = 1;
    }
   
    # copy input sequence file into results directory
    my $cmd2 = "apptainer exec -C -B $host_results_rootdir --pwd $container_results_dir --overlay $overlay_file $sif_file bash -c \"cp $host_seqfile $container_results_dir/\" 2>> psortb.err";
    if ($verbose) {
      print "Running: $cmd2\n\n";
    }
    system($cmd2);

    # run psortb command
    my $cmd3 = "apptainer exec -C -B $host_results_rootdir --pwd $container_results_dir --env MOUNT=$host_results_dir --overlay $overlay_file $sif_file bash -c \"ln -s /results /tmp/results; /usr/local/psortb/bin/psort $extra_args -i $container_seqfile\" 2>> psortb.err";
    if ($verbose) {
      print "Running: $cmd3\n\n";
    }
    system($cmd3);

    # copy results from overlay to host file system
    my $cmd4 = "apptainer exec -C -B $host_results_rootdir --pwd $container_results_dir --overlay $overlay_file $sif_file bash -c \"cp -rp $container_results_dir/* $host_results_dir/\" 2>> psortb.err";
    if ($verbose) {
      print "\nRunning: $cmd4\n\n";
    }
    system($cmd4);

    # remove copied files (if they were created for the purposes of this workflow)
    delete_file($host_seqfile, $seqfile_copied);
    delete_file($overlay_file, $overlay_created_flag);
}

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

    my $cmd = "cp $orig_file $new_file";
    system($cmd);
    if ($@) {
      die "Could not copy file: $cmd: $@\n";
    }
    return(1);
}

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 (host results dir) because inside container - 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>)
    # Also remove container options
    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') ||
            ($$args[$i] eq '-s') ||
            ($$args[$i] eq '--sif-file') ||
            ($$args[$i] eq '--overlay-file') ||
            ($$args[$i] eq '--overlay-size-mb')) 
        {
            $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\n");
    print("Required options:\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\n");
    print("Extra PSORTb options:\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("  --x-skip-localization  Comma-separated list of localizations to skip\n\n");
    print("Container options:\n");
    print("  --sif_file        Apptainer image file (sif file) location (default: ./psortb.sif)\n");
    print("  --overlay-file    Specify an overlay image file you want to use.\n");
    print("                    (Default: results.img created and destroyed after\n"); 
    print("                    PSORTb analysis ends)\n");
    print("  --overlay-size-mb Maximum size of results folder (if it doesn't \n");
    print("                    already exist) in megabytes (default: 300)\n\n");
    print("Other options:\n");
    print("  --verbose, -v     Be verbose while running\n");
    print "  --version         Print the version of PSORTb\n";    
    print("  --help, -h        Displays usage information\n\n");

    exit(0);
}

main();