view release on metacpan or  search on metacpan

lib/Bioinfo/App/Cmd/Fasta/Cmd/Split.pm  view on Meta::CPAN

package Bioinfo::App::Cmd::Fasta::Cmd::Split;
use Modern::Perl;
use Moo;
use MooX::Cmd;
use MooX::Options prefer_commandline => 1;
use IO::All;

our $VERSION = '0.1.15'; # VERSION: 
# ABSTRACT: my perl module and CLIs for Biology


option input => (
  is  => 'ro',
  required  => 1,
  format  => 's',
  short => 'i',
  doc => 'a file of fasta format'
);


option num => (
  is  => 'ro',
  format  => 'i',
  short => 'n',
  doc => ''
);


option outdir => (
  is => 'ro',
  format => 's',
  short => 'o',
  doc => '',
);


option prefix => (
  is => 'ro',
  format => 's',
  short => 'p',
  default => sub { '' },
  doc => 'the prefix of the split file',
);


sub execute {
  my ($self, $args_ref, $chain_ref) = @_;
  $self->options_usage unless (@$args_ref);
  my $input = $self->input;
  my $outdir = $self->outdir;
  my $num = $self->num;
  my $prefix = $self->prefix;
  system("mkdir -p $outdir") unless -e $outdir;
  my %id2seq;
  my $io_fa = io($input)->chomp;
  my ($total_base, $seqid) = (0, '');
  while (defined (my $line = $io_fa->getline)) {
    if ($line =~/^>/) {
      $seqid = $line;
    } else {
      $id2seq{$seqid} .= $line;
      $total_base += length($line);