scripts/taxa/bp_local_taxonomydb_query.pl

   1 #!/usr/bin/perl
   2 use strict;
   3 use warnings;
   4 use Bio::DB::Taxonomy;
   5
   6 use strict;
   7 use Getopt::Long;
   8 my $verbose = 0;
   9 my $plain   = 0;
  10 my ($nodesfile,$namesfile);
  11 my $idx_dir = '/tmp/idx';
  12 GetOptions('v|verbose' => \$verbose,
  13            'nodes:s'   => \$nodesfile,
  14            'names:s'   => \$namesfile,
  15            'idx:s'     => \$idx_dir,
  16            'h|help'    => sub{ exec('perldoc',$0);
  17                                 exit(0)
  18                                 } );
  19
  20 unless( @ARGV || $nodesfile || $namesfile ) {
  21     exec('perldoc',$0);
  22     exit(0);
  23 }
  24 mkdir($idx_dir) unless -d $idx_dir;
  25
  26 my $db = new Bio::DB::Taxonomy(-source    => 'flatfile',
  27                                -nodesfile => $nodesfile,
  28                                -namesfile => $namesfile,
  29                                -directory => $idx_dir);
  30 foreach my $sp ( @ARGV ) {
  31     my $node = $db->get_Taxonomy_Node(-name => $sp);
  32     if( defined $node ) {
  33         print "id is ", $node->id, "\n"; # 9606
  34         print "rank is ", $node->rank, "\n"; # species
  35         print "scientific name is ", $node->scientific_name, "\n"; # Homo sapiens
  36         print "division is ", $node->division, "\n"; # Primates
  37     } else {
  38         warn("no node found for query $sp");
  39     }
  40 }
  41
  42 =head1 NAME
  43
  44 bp_local_taxonomydb_query - query a local TaxonomyDB for species or taxonid
  45
  46 =head1 DESCRIPTION
  47
  48 This script provides an example implementation of access to a local
  49 Taxonomy database implemented with Berkeley DB (DB_File module is needed).
  50
  51 Usage:
  52
  53  bp_local_taxonomydb_query.PLS: [-v] --nodes nodes.dmp --names names.dmp "Genus1 species1" "Genus2 species2"
  54
  55 Providing the nodes.dmp and names.dmp files from the NCBI Taxonomy
  56 dump (see Bio::DB::Taxonomy::flatfile for more info) is only necessary
  57 on the first time running.  This will create the local indexes and may
  58 take quite a long time.  However once created, these indexes will
  59 allow fast access for species to taxon id OR taxon id to species name
  60 lookups.