view release on metacpan or  search on metacpan

lib/App/Rangeops/Command/clean.pm  view on Meta::CPAN

package App::Rangeops::Command::clean;
use strict;
use warnings;
use autodie;

use App::Rangeops -command;
use App::Rangeops::Common;

sub abstract {
    return 'replace ranges within links, incorporate hit strands and remove nested links';
}

sub opt_spec {
    return (
        [   "replace|r=s",
            "Two-column tsv file, normally produced by command merge."
        ],
        [   "bundle|b=i",
            "Bundle overlapped links. This value is the overlapping size, default is [0]. Suggested value is [500].",
            { default => 0 },
        ],
        [ "outfile|o=s", "Output filename. [stdout] for screen." ],
        [ "verbose|v",   "Verbose mode.", ],
    );
}

sub usage_desc {
    return "rangeops clean [options] <infiles>";
}

sub description {
    my $desc;
    $desc .= ucfirst(abstract) . ".\n";
    $desc
        .= "\t<infiles> are bilateral links files, with or without hit strands\n";
    return $desc;
}

sub validate_args {
    my ( $self, $opt, $args ) = @_;

    if ( !@{$args} ) {
        my $message = "This command need one or more input files.\n\tIt found";
        $message .= sprintf " [%s]", $_ for @{$args};
        $message .= ".\n";
        $self->usage_error($message);
    }
    for ( @{$args} ) {
        next if lc $_ eq "stdin";
        if ( !Path::Tiny::path($_)->is_file ) {
            $self->usage_error("The input file [$_] doesn't exist.");
        }
    }

    if ( !exists $opt->{outfile} ) {
        $opt->{outfile}
            = Path::Tiny::path( $args->[0] )->absolute . ".clean.tsv";
    }
}

sub execute {
    my ( $self, $opt, $args ) = @_;

    #----------------------------#
    # Load replaces
    #----------------------------#
    my $info_of    = {};    # info of ranges
    my $replace_of = {};
    if ( $opt->{replace} ) {
        print STDERR "==> Load replaces\n" if $opt->{verbose};
        for my $line ( App::RL::Common::read_lines( $opt->{replace} ) ) {
            $info_of = App::Rangeops::Common::build_info( [$line], $info_of );

            my @parts = split /\t/, $line;
            if ( @parts == 2 ) {
                $replace_of->{ $parts[0] } = $parts[1];
            }
        }
    }

    #----------------------------#
    # Replacing and incorporating
    #----------------------------#
    print STDERR "==> Incorporating strands\n" if $opt->{verbose};
    my @lines;
    for my $file ( @{$args} ) {
        for my $line ( App::RL::Common::read_lines($file) ) {
            $info_of = App::Rangeops::Common::build_info( [$line], $info_of );

            my @new_parts;

            # replacing
            for my $part ( split /\t/, $line ) {

                if ( exists $replace_of->{$part} ) {
                    my $original = $part;
                    my $replaced = $replace_of->{$part};

                    # create new info, don't touch anything of $info_of