view release on metacpan or  search on metacpan

lib/App/RL/Command/stat2.pm  view on Meta::CPAN

package App::RL::Command::stat2;
use strict;
use warnings;
use autodie;

use App::RL -command;
use App::RL::Common;

sub abstract {
    return 'coverage statistics on another runlist for runlists';
}

sub opt_spec {
    return (
        [ "outfile|o=s", "output filename. [stdout] for screen" ],
        [ "op=s",     "operations: intersect, union, diff or xor", { default  => "intersect" } ],
        [ "size|s=s", "chr.sizes",                                 { required => 1 } ],
        [ "base|b=s", "basename of infile2", ],
        [ "remove|r", "remove 'chr0' from chromosome names" ],
        [ "mk",       "first YAML file contains multiple sets of runlists" ],
        [ "all",      "only write whole genome stats" ],
        { show_defaults => 1, }
    );
}

sub usage_desc {
    return "runlist stat2 [options] <infile1> <infile2>";
}

sub description {
    my $desc;
    $desc .= ucfirst(abstract) . ".\n";
    return $desc;
}

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

    if ( @{$args} != 2 ) {
        my $message = "This command need two 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 ( $opt->{op} =~ /^dif/i ) {
        $opt->{op} = 'diff';
    }
    elsif ( $opt->{op} =~ /^uni/i ) {
        $opt->{op} = 'union';
    }
    elsif ( $opt->{op} =~ /^int/i ) {
        $opt->{op} = 'intersect';
    }
    elsif ( $opt->{op} =~ /^xor/i ) {
        $opt->{op} = 'xor';
    }
    else {
        Carp::confess "[@{[$opt->{op}]}] invalid\n";
    }

    if ( !exists $opt->{base} ) {
        $opt->{base} = Path::Tiny::path( $args->[1] )->basename( ".yaml", ".yml" );
    }

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

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

    #----------------------------#
    # Loading
    #----------------------------#
    my $length_of = App::RL::Common::read_sizes( $opt->{size}, $opt->{remove} );

    # file1
    my $s1_of = {};
    my @keys;
    if ( $opt->{mk} ) {
        my $yml = YAML::Syck::LoadFile( $args->[0] );
        @keys = sort keys %{$yml};

        for my $key (@keys) {
            $s1_of->{$key} = App::RL::Common::runlist2set( $yml->{$key}, $opt->{remove} );
        }
    }
    else {