view release on metacpan or  search on metacpan

lib/App/Fasops/Command/links.pm  view on Meta::CPAN

package App::Fasops::Command::links;
use strict;
use warnings;
use autodie;

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

sub abstract {
    return 'scan blocked fasta files and output bi/multi-lateral range links';
}

sub opt_spec {
    return (
        [ "outfile|o=s", "Output filename. [stdout] for screen." ],
        [ "pair|p",      "pairwise links" ],
        [ "best|b",      "best-to-best pairwise links" ],
        { show_defaults => 1, }
    );
}

sub usage_desc {
    return "fasops links [options] <infile> [more infiles]";
}

sub description {
    my $desc;
    $desc .= ucfirst(abstract) . ".\n";
    $desc .= <<'MARKDOWN';

* <infiles> are paths to axt files, .axt.gz is supported
* infile == stdin means reading from STDIN

MARKDOWN

    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 . ".tsv";
    }
}

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

    my @links;
    for my $infile ( @{$args} ) {
        my $in_fh;
        if ( lc $infile eq "stdin" ) {
            $in_fh = *STDIN{IO};
        }
        else {
            $in_fh = IO::Zlib->new( $infile, "rb" );
        }

        my $content = '';    # content of one block
        while (1) {
            last if $in_fh->eof and $content eq '';
            my $line = '';
            if ( !$in_fh->eof ) {
                $line = $in_fh->getline;
            }
            next if substr( $line, 0, 1 ) eq "#";

            if ( ( $line eq '' or $line =~ /^\s+$/ ) and $content ne '' ) {
                my $info_of = App::Fasops::Common::parse_block( $content, 1 );
                $content = '';

                my @headers = keys %{$info_of};

                if ( $opt->{best} ) {
                    my @matrix = map { [ (undef) x ( scalar @headers ) ] } 0 .. $#headers;

                    # distance is 0 for same sequence
                    for my $i ( 0 .. $#headers ) {
                        $matrix[$i][$i] = 0;
                    }

                    # compute a triangle, fill full matrix
                    for ( my $i = 0; $i <= $#headers; $i++ ) {
                        for ( my $j = $i + 1; $j <= $#headers; $j++ ) {
                            my $D = App::Fasops::Common::pair_D(