db/0001_load_tomato_gen_pubs.pl

   1 #!/usr/bin/env perl
   2
   3
   4 =head1 NAME
   5
   6 [ this script name ].pl
   7
   8 =head1 SYNOPSIS
   9
  10   this_script.pl [options]
  11
  12   Options:
  13
  14     -D <dbname> (mandatory)
  15       dbname to load into
  16
  17     -H <dbhost> (mandatory)
  18       dbhost to load into
  19
  20     -p <script_executor_user> (mandatory)
  21       username to run the script
  22
  23     -F force to run this script and don't stop it by
  24        missing previous db_patches
  25
  26   Note: If the first time that you run this script, obviously
  27         you have no previous dbversion row in the md_dbversion
  28         table, so you need to force the execution of this script
  29         using -F
  30
  31 =head1 DESCRIPTION
  32
  33 store in the database the tomato genome publications as pubprops
  34
  35 =head1 AUTHORS
  36
  37 Jean Lee <jhl233@cornell.edu>
  38 Naama Menda < nm249@cornell.edu>
  39
  40 =head1 COPYRIGHT & LICENSE
  41
  42 Copyright 2010 Boyce Thompson Institute for Plant Research
  43
  44 This program is free software; you can redistribute it and/or modify
  45 it under the same terms as Perl itself.
  46
  47 =cut
  48
  49
  50 use strict;
  51 use warnings;
  52
  53 use Bio::Chado::Schema;
  54 use Pod::Usage;
  55 use Getopt::Std;
  56 use CXGN::DB::InsertDBH;
  57 use CXGN::Metadata::Dbversion;   ### Module to interact with the metadata.md_dbversion table
  58
  59
  60 ## Declaration of the parameters used to run the script
  61
  62 our ($opt_H, $opt_D, $opt_p, $opt_F, $opt_h);
  63 getopts("H:D:p:Fh");
  64
  65 ## If is used -h <help> or none parameters is detailed print pod
  66
  67 if (!$opt_H && !$opt_D && !$opt_p && !$opt_F && !$opt_h) {
  68     print STDOUT "No optionas passed. Printing help\n\n";
  69     pod2usage(1);
  70 }
  71 elsif ($opt_h) {
  72     pod2usage(1);
  73 }
  74
  75
  76 ## Declaration of the name of the script and the description
  77
  78 my $patch_name = '0001_load_tomato_gen_pubs.pl';
  79 my $patch_descr = 'This script stores pubprop for the tomato genome publications. It assumes these are ALREADY STORED in the database. The best way to load first the publications is by using the web interface ';
  80
  81 print STDOUT "\n+--------------------------------------------------------------------------------------------------+\n";
  82 print STDOUT "Executing the patch:\n   $patch_name.\n\nDescription:\n  $patch_descr.\n\nExecuted by:\n  $opt_p.";
  83 print STDOUT "\n+--------------------------------------------------------------------------------------------------+\n\n";
  84
  85 ## And the requeriments if you want not use all
  86 ##
  87 my @previous_requested_patches = (   ## ADD HERE
  88    );
  89
  90 ## Specify the mandatory parameters
  91
  92 if (!$opt_H || !$opt_D) {
  93     print STDOUT "\nMANDATORY PARAMETER ERROR: -D <db_name> or/and -H <db_host> parameters has not been specified for $patch_name.\n";
  94 }
  95
  96 if (!$opt_p) {
  97     print STDOUT "\nMANDATORY PARAMETER ERROR: -p <script_executor_user> parameter has not been specified for $patch_name.\n";
  98 }
  99
 100 ## Create the $schema object for the db_version object
 101 ## This should be replace for CXGN::DB::DBICFactory as soon as it can use CXGN::DB::InsertDBH
 102
 103 my $db =  CXGN::DB::InsertDBH->new(
 104                                      {
 105                                          dbname => $opt_D,
 106                                          dbhost => $opt_H
 107                                      }
 108                                    );
 109 my $dbh = $db->get_actual_dbh();
 110
 111 print STDOUT "\nCreating the Metadata Schema object.\n";
 112
 113 my $metadata_schema = CXGN::Metadata::Schema->connect(
 114                                                        sub { $dbh },
 115                                                       { on_connect_do => ['SET search_path TO metadata;'] },
 116                                                       );
 117
 118 print STDOUT "\nChecking if this db_patch was executed before or if previous db_patches have been executed.\n";
 119
 120 ### Now it will check if you have runned this patch or the previous patches
 121
 122 my $dbversion = CXGN::Metadata::Dbversion->new($metadata_schema)
 123                                          ->complete_checking( {
 124                                                                  patch_name  => $patch_name,
 125                                                                  patch_descr => $patch_descr,
 126                                                                  prepatch_req => \@previous_requested_patches,
 127                                                                  force => $opt_F
 128                                                               }
 129                                                              );
 130
 131
 132 ### CREATE AN METADATA OBJECT and a new metadata_id in the database for this data
 133
 134 my $metadata = CXGN::Metadata::Metadbdata->new($metadata_schema, $opt_p);
 135
 136 ### Get a new metadata_id (if you are using store function you only need to supply $metadbdata object)
 137
 138 my $metadata_id = $metadata->store()
 139                            ->get_metadata_id();
 140
 141 ### Now you can insert the data using different options:
 142 ##
 143 ##  1- By sql queryes using $dbh->do(<<EOSQL); and detailing in the tag the queries
 144 ##
 145 ##  2- Using objects with the store function
 146 ##
 147 ##  3- Using DBIx::Class first level objects
 148 ##
 149
 150 ## In this case we will use the SQL tag
 151
 152 print STDERR "\nExecuting the SQL commands.\n";
 153
 154
 155 my $schema = Bio::Chado::Schema->connect(sub { $db->get_actual_dbh() } ,
 156     { on_connect_do => ['SET SEARCH_PATH TO public, metadata;'] } );
 157
 158 #---arrays of pubmed ids and titles of publications---
 159 my @pmid = (
 160     18254380,
 161     18317508,
 162     17565940,
 163     16830097,
 164     16524981,
 165     16489216,
 166     16208505,
 167     16010005,
 168     10645957,
 169     10382301,
 170     10224272,
 171     18469880,
 172     8662247,
 173     8653264,
 174     8647403,
 175 );
 176
 177 my @titles = ('A Snapshot of the Emerging Tomato Genome Sequence', 'Estimation of nuclear DNA content of plants by flow cytometry');
 178
 179
 180 #---add pubmed publications to pubprop---
 181 my $pmdb = $schema->resultset('General::Db')->find({name => 'PMID'})
 182     or die "no db found with name 'PMID'";
 183
 184 foreach my $item ( @pmid ) {
 185     my $dbxref = $pmdb->find_related(
 186         'dbxrefs',
 187         {accession => $item}
 188        )
 189         or die "no dbxref found for PMID $item";
 190
 191     my $pub = $dbxref->find_related('pub_dbxrefs', {})
 192                      ->find_related('pub', {})
 193                      ->create_pubprops(
 194                          {'tomato genome publication' => '1'},
 195                          { autocreate => 1 }
 196                         );
 197 }
 198
 199 #---add other publications manually---
 200 foreach my $title ( @titles ) {
 201     my $pub = $schema->resultset( "Pub::Pub" )->find({ title => $title })
 202         or die "no publication found with title '$title'";
 203
 204     $pub->create_pubprops(
 205         {'tomato genome publication' => '1'},
 206         {autocreate => 1}
 207        );
 208 }
 209
 210 ## Now it will add this new patch information to the md_version table.  It did the dbversion object before and
 211 ## set the patch_name and the patch_description, so it only need to store it.
 212
 213
 214 $dbversion->store($metadata);
 215
 216 print STDOUT "DONE!\n";
 217
 218 $dbh->commit;
 219
 220 __END__
 221