view release on metacpan or  search on metacpan

lib/Bio/App/SELEX/RNAmotifAnalysis.pm  view on Meta::CPAN

#!/usr/bin/env perl
package Bio::App::SELEX::RNAmotifAnalysis;
# ABSTRACT: Cluster SELEX sequences and calculate their structures
use 5.008;
use strict;
use warnings;
use Text::LevenshteinXS qw( distance );
use Config::Tiny;
use autodie;
use Hash::Util qw( lock_keys );
use List::Util qw( min );
use Getopt::Long;
use Carp qw( croak confess);

my $DEFAULT_CONFIG = 'cluster.cfg';

# CONSTANTS
my $TRUE         = 1;
my $FALSE        = 0;
my $SPACE        = q{ };
my $EMPTY_STRING = q{};
my $VERBOSE      = 1;

my $FASTQ_TYPE  = 'fastq';
my $SIMPLE_TYPE = 'simple';

# Act like a script if called as one
unless ( caller() ) { main(); }

sub main {

    my $max_clusters    = 10;
    my $num_cpus        = 5;
    my $max_distance    = 5;
    my $max_top_seqs    = 300;
    my $config_filename = $DEFAULT_CONFIG;
    my $options         = GetOptions(

        # Required (one of these)
        "$FASTQ_TYPE=s"        => \my $fastq,
        "$SIMPLE_TYPE=s"       => \my $simple,

        # Optional
        'max_distance=i' => \$max_distance,
        'max_clusters=i' => \$max_clusters,
        'max_top_seqs=i' => \$max_top_seqs,
        'seed=s'         => \my $seed_filename,
        'cpus=i'         => \$num_cpus,
        'config=s'       => \$config_filename,
        'run'            => \my $run_scripts,
    );

    my $file_type;
    my $infile;
    if(defined $fastq){
        if(defined $simple){
            warn "--$FASTQ_TYPE and --$SIMPLE_TYPE flags are mutually exclusive!\n";
            help();
        }
        $infile    = $fastq;
        $file_type = $FASTQ_TYPE;
    }elsif(defined $simple){
        $infile    = $simple;
        $file_type = $SIMPLE_TYPE;
    }else{
        warn "either --$FASTQ_TYPE or --$SIMPLE_TYPE must be used!\n";
        help();
    }

    # Eliminate case sensitivity for 'type'
    $file_type = lc $file_type;

    my $config = get_config($config_filename);

    open( my $fh_in, '<', $infile );

    my $seed_fh;
    if ( defined $seed_filename && -e $seed_filename ) {
        open( $seed_fh, '<', $seed_filename );
    }

    my ($cluster_href, $distance_href) = cluster(
        fh           => $fh_in,
        max_distance => $max_distance,
        max_clusters => $max_clusters,
        seed_fh      => $seed_fh,
        file_type    => $file_type,
    );