examples/db/get_seqs.pl

   1 #!/usr/bin/perl
   2 use strict;
   3 use vars qw($USAGE);
   4 use Carp;
   5 use Getopt::Long;
   6 use Bio::SeqIO;
   7
   8 $USAGE = "get_seqs.pl\t[--db=DBNAME] [--format=FORMAT] \n\t\t[--output=FILENAME] [--proxy=PROXY] accession1, accession2, ...\n Defaults: db=GenBank format=fasta output=STDOUT proxy=none\n See LWP::UserAgent for more information on proxy syntax";
   9 my %dbs = (
  10   'genbank'   => 'Bio::DB::GenBank',
  11   'embl'      => 'Bio::DB::EMBL',
  12   'swissprot' => 'Bio::DB::SwissProt',
  13 );
  14
  15 my ($db,$format,$file,$proxy,$help) = ( 'genbank', 'fasta' );
  16
  17 &GetOptions
  18     (
  19      'db:s'       => \$db,
  20      'f|format:s' => \$format,
  21      "file|out|output:s" => \$file,
  22      'proxy:s'           => \$proxy,
  23      "h|\?|help"  => \$help ,
  24      );
  25
  26 if( $help ) { print $USAGE, "\n";exit; }
  27
  28 if( $db =~ /gb|gen|genbank/i ) {
  29     $db = 'genbank';
  30 } elsif( $db =~ /embl|em|e/i ) {
  31     $db = 'embl';
  32 } elsif( $db =~ /swiss|sp/i ) {
  33     $db = 'swissprot';
  34 } else {
  35     croak("Unknown db parameter '$db' valid parameters are (" . join(',', keys %dbs) . ")");
  36 }
  37
  38 my %params = ( '-format' => $format );
  39
  40 if( defined $file ) {
  41     $params{'-file'} = ">$file";
  42 } else {
  43     $params{'-fh'} = \*STDOUT;
  44 }
  45
  46 my $seqio = new Bio::SeqIO(%params);
  47
  48 my $remotedb;
  49
  50 eval {
  51     my $filename = "$dbs{$db}.pm";
  52     $filename =~ s!::!/!g;
  53     require $filename;
  54     $remotedb = "$dbs{$db}"->new();
  55 };
  56
  57 die($@) unless ! $@;
  58
  59 if( defined $proxy ) { $remotedb->proxy($proxy); }
  60
  61 my $stream;
  62
  63 if( $remotedb->can('get_Stream_by_batch') ) {
  64     $stream = $remotedb->get_Stream_by_batch(@ARGV);
  65 } else {
  66     $stream = $remotedb->get_Stream_by_acc(\@ARGV);
  67 }
  68
  69 while( my $seq = $stream->next_seq ) {
  70     $seqio->write_seq($seq);
  71 }