bin/load_pub_doi.pl

   1 #!/usr/bin/perl
   2
   3 use Getopt::Std;
   4 use CXGN::DB::InsertDBH;
   5 use Bio::Chado::Schema;
   6 use Data::Dumper;
   7 use CXGN::Chado::Publication;
   8 use CXGN::Tools::Pubmed;
   9 use strict;
  10 use warnings;
  11
  12 use vars qw | $opt_H $opt_D |;
  13
  14 getopts('H:D:');
  15
  16 my $dbh = CXGN::DB::InsertDBH->new( {
  17     dbhost => $opt_H,
  18     dbname => $opt_D,
  19     } );
  20
  21
  22
  23 my $schema = Bio::Chado::Schema->connect( sub { $dbh->get_actual_dbh() }, { on_connect_do => ['SET search_path TO public;'] , autocommit => 1 } );
  24
  25 print STDERR "finding dbxrefs for all publications\n";
  26 my $dbxref_rs = $schema->resultset("General::Dbxref")->search(
  27     {
  28         'db.name' => 'PMID',
  29     },
  30     {
  31         join       => [ 'db', 'pub_dbxrefs' ] ,
  32         "+select"  => [ "pub_dbxrefs.pub_id" ],
  33         "+as"      => [ "pub_id" ],
  34     }
  35     );
  36 my $pub_count;
  37 my $doi_count;
  38 while (my $dbxref = $dbxref_rs->next ) {
  39     $pub_count++;
  40     my $accession = $dbxref->accession;
  41     my $pub_id = $dbxref->get_column("pub_id");
  42     if (!$pub_id) {
  43         warn "No pub_id exists for accession $accession! Skipping ! \n\n";
  44         next;
  45     }
  46     my $pub = CXGN::Chado::Publication->new( $dbh, $pub_id );
  47     $pub->set_accession($accession);
  48     my $pubmed = CXGN::Tools::Pubmed->new($pub);
  49
  50     my $eid = $pub->get_eid;
  51     my $title = $pub->get_title;
  52     if ( $eid ) {
  53         $doi_count++;
  54         print STDERR "Found DOI $eid\n";
  55         my $db = $schema->resultset("General::Db")->find_or_create(
  56             {
  57                 name       => 'DOI',
  58                 urlprefix =>  'http://',
  59                 url        => 'doi.org',
  60             } );
  61
  62         my $e_dbxref = $db->find_or_create_related("dbxrefs" , { accession => $eid } );
  63         $e_dbxref->find_or_create_related("pub_dbxrefs", { pub_id => $pub_id } ) ;
  64
  65         print STDERR "Loaded DOI $eid for publication $pub_id (accession $accession) dbxref id = " . $e_dbxref->get_column('dbxref_id') . " **\n";
  66     } else {
  67         print STDERR "no DOI for pub_id $pub_id , pubmed accession = $accession\n";
  68     }
  69 }
  70 $schema->txn_commit;
  71 print STDERR "\nDONE. Found $pub_count publications, $doi_count DOIs \n\n";