scripts/taxa/bp_query_entrez_taxa.pl

   1 #!/usr/bin/perl
   2 # This is a -*-Perl-* file (make my emacs happy)
   3
   4 =head1 NAME
   5
   6 bp_query_entrez_taxa - query Entrez taxonomy database and print out information
   7
   8 =head1 USAGE
   9
  10 bp_query_entrez_taxa "Homo sapiens" "Saccharomyces cerevisiae" Rhizopus Metazoa
  11 bp_query_entrez_taxa -gi 28800981 -gi 54301680 -db nucleotide
  12 bp_query_entrez_taxa -gi 71836523 -db protein
  13
  14  Provide the genus and species name in quotes, you can also query for
  15  a non-species node like Family or Order
  16
  17 Command-line options:
  18    -v or --verbose  : print verbose debugging info
  19    -gi              : one or many GI numbers to lookup taxon id for
  20    -db              : the sequence db (nucleotide or protein) the GI is for
  21
  22    other arguments are assumed to be species names to lookup in taxonomy db
  23
  24
  25 =head1 AUTHOR
  26
  27 Jason Stajich jason-at-bioperl-dot-org
  28
  29 =cut
  30
  31 use strict;
  32 use warnings;
  33 use Bio::DB::Taxonomy;
  34 use Getopt::Long;
  35
  36 my $verbose = 0;
  37 my (@gi, $dbname);
  38 GetOptions('v|verbose' => \$verbose,
  39            'gi:i'      => \@gi,
  40            'db:s'      => \$dbname);
  41
  42 my $db = new Bio::DB::Taxonomy(-source => 'entrez', -verbose => $verbose);
  43 if( @gi ) {
  44     my @nodes= $db->get_Taxonomy_Node(-gi => \@gi,
  45                                       -db => $dbname);
  46     for my $node ( @nodes ) {
  47         my $gi = shift @gi;
  48         print " for gi $gi:\n";
  49         print " taxonid is ",$node->ncbi_taxid,"\n";
  50         print " node is ", join(", ",$node->classification), "\n";
  51         print " species is ", $node->species,"\n";
  52         print " parent is ", $node->parent_id, "\n";
  53         print " rank is ", $node->rank, "\n";
  54         print " genetic_code  ", $node->genetic_code, "\n";
  55         print " mito_genetic_code  ", $node->mitochondrial_genetic_code, "\n";
  56         print " scientfic name is ", $node->binomial, "\n";
  57     }
  58 }
  59
  60 print "\n\n";
  61 for my $name ( @ARGV ) {
  62     my $taxonid = $db->get_taxonid($name);
  63     my $node   = $db->get_Taxonomy_Node(-taxonid => $taxonid);
  64     print "taxonid is $taxonid\n";
  65
  66     print " node is ", join(", ",$node->classification), "\n";
  67     print " species is ", $node->species,"\n";
  68     print " parent is ", $node->parent_id, "\n";
  69     print " rank is ", $node->rank, "\n";
  70     print " genetic_code  ", $node->genetic_code, "\n";
  71     print " mito_genetic_code  ", $node->mitochondrial_genetic_code, "\n";
  72     print " scientfic name is ", $node->binomial, "\n";
  73     print " common name is ", $node->common_name, "\n";
  74     print " create date is ", $node->create_date, "\n";
  75     print " update date is ", $node->update_date, "\n";
  76     print " pub date is ", ($node->pub_date || ''), "\n";
  77     print " variant is ", $node->variant, "\n";
  78     print " sub_species is ", $node->sub_species, "\n";
  79     print " organelle is ", $node->organelle, "\n";
  80     print " division is ", $node->division, "\n";
  81 }