scripts/DB/bp_biogetseq.pl

   1 #!perl
   2 #
   3 # OBDA Registry compliant sequence retrieval script
   4 #
   5 # Copyright Heikki Lehvaslaiho <heikki-at-bioperl-dot-org>
   6 # You may distribute this program under the same terms as perl itself
   7
   8
   9 use Bio::DB::Registry;
  10 use Bio::SeqIO;
  11 use Getopt::Long;
  12 use strict;
  13 use warnings;
  14
  15 my ($help, $format, $namespace, $dbname) = ('', 'embl', 'acc', 'embl');
  16 GetOptions ("help" => \$help, "format=s" => \$format,
  17             "namespace=s" => \$namespace, "dbname=s" => \$dbname );
  18 if ($help || !@ARGV) {
  19   system("perldoc $0");
  20   exit 0;
  21 }
  22
  23 my $get_function = 'get_Seq_by_'. $namespace;
  24
  25 my $registry = new Bio::DB::Registry();
  26 while (my $id = shift) {
  27     my $db = $registry->get_database($dbname);
  28     my $seq = $db->$get_function($id);
  29
  30     if ($seq) {
  31         my $out = new Bio::SeqIO('-format' => $format);
  32         $out->write_seq($seq);
  33     } else {
  34         print STDERR "Could not find sequence with identifier [$id]\n";
  35     }
  36 }
  37
  38 =head1 NAME
  39
  40 bp_biogetseq - sequence retrieval using OBDA registry
  41
  42 =head1 DESCRIPTION
  43
  44 This script retrives sequences from the source defined by users
  45 registry setup.  The current alternatives are from a local indexed
  46 file, sql database or over the web.
  47
  48 =head1 USAGE
  49
  50   Usage: bp_biogetseq --dbname embl --format embl
  51                    --namespace acc id [ ids... ]
  52          * dbname defaults to embl
  53          * format defaults to embl
  54          * namespace defaults to 'acc' ['id', 'acc', 'version']
  55          * unnamed arguments are ids in the given namespace
  56
  57 =cut
  58