bin/download_obo.pl

   1
   2 =head1 NAME
   3
   4 download_obo.pl - script to download obo file of an ontology
   5
   6 =head1 SYNOPSIS
   7
   8 perl download_obo.pl -i prefix -H host -D dbname
   9
  10 =head1 DESCRIPTION
  11
  12 Download an obo file from your database. The file will contain all terms listed under the provided ontology prefix (db.name).
  13 The obo filename is $prefix_breedbase.obo.
  14 After downloading the file it is recommended to test it in the Protege program to make sure it is a valid obo,
  15 and loading back into the database with gmod_load_cvterms.pl  <see the solgenomics/chado_tools repo>
  16
  17
  18 This script currently prints the following fields:
  19
  20     [Term]
  21     id: $prefix:$accession
  22     name: $cvterm_name
  23     namespace: $namespace
  24     def: "$def" #if the cvterm 'definition' field is populated
  25     is_obsolete: true #if cvterm is_obsolete field is 'true'
  26     synonym: $synonym_name []  # one row for each synonym
  27     xref: $xref_cvterm [] # one row for each cvterm xref that is not for the cvterm is_for_definition
  28     is_a: $cvterm # one row for each is_a relationship object
  29     relationship: $typedef $cvterm # one row for each relationship type that is not is_a (e.g. variable_of, method_of, scale_of)
  30
  31     --------------------------------------------
  32
  33     Example:
  34
  35     [Term]
  36     id: CO_334:0000009
  37     name: initial vigor assessment 1-7
  38     namespace: cassava_trait
  39     def: "Visual assessment of plant vigor during establishment scored one month after planting. 3 = Not vigorous, 5 = Medium vigor, 7 = highly vigorous." [CO:curators]
  40     synonym: "Can't fall when there is strong wind" EXACT []
  41     synonym: "IVig_IITAVisScg_1to7" EXACT []
  42     synonym: "vigor" EXACT []
  43     xref: TO:0000250
  44     is_a: CO_334:0001000 ! Variables
  45     is_a: CO_334:0002010 ! Farmer trait
  46     relationship: variable_of CO_334:0000386 ! Initial vigor
  47     relationship: variable_of CO_334:0010228 ! Visual Rating: Initial vigor_method
  48     relationship: variable_of CO_334:0100434 ! 7pt scale
  49
  50     ------------------------------------------
  51
  52 If there are any cvterm.is_relationshiptype for this ontology they will be printed as
  53     [Typedef]
  54     id: $cvterm_name
  55     name: $cvterm_name
  56
  57
  58 =head1 COMMAND-LINE OPTIONS
  59
  60     -H host name
  61     -D database name
  62     -i prefix for the ontology (e.g. CO_334)
  63     -t Test run. Rolls back at the end.
  64
  65 =head1 AUTHOR
  66
  67 Naama Menda <nm249@cornell.edu>
  68
  69 =cut
  70
  71 use strict;
  72
  73 use Getopt::Std;
  74 use Data::Dumper;
  75 use Try::Tiny;
  76 use DateTime;
  77
  78 use Bio::Chado::Schema;
  79 use CXGN::DB::InsertDBH;
  80 use File::Slurp;
  81
  82 our ($opt_H, $opt_D, $opt_U, $opt_P, $opt_i);
  83
  84 getopts('H:D:U:P:i:');
  85
  86 my $dbhost = $opt_H;
  87 my $dbname = $opt_D;
  88 my $dbuser = $opt_U;
  89 my $dbpass = $opt_P;
  90 my $prefix = $opt_i;
  91
  92         dbname=>$dbname,
  93         dbargs => {AutoCommit => 0,
  94         RaiseError => 1}
  95     } );
  96
  97 print STDERR "Connecting to database...\n";
  98 my $schema= Bio::Chado::Schema->connect(  sub { $dbh->get_actual_dbh() } );
  99
 100 my $obo_file = $prefix . ".breedbase.obo";
 101
 102 #resultset of all cvterms
 103 my $cvterm_rs = $schema->resultset("Cv::Cvterm")->search(
 104     { 'db.name' => $prefix },
 105     { join => [ 'cv', { dbxref => 'db' } ]  ,
 106     '+select' => ['cv.name', 'dbxref.accession'],
 107     '+as' => [ 'cv_name', 'dbxref_accession' ]
 108 } );
 109
 110 my $date = DateTime->now();
 111
 112 my $obo_header = "format-version: 1.2
 113 ontology: $prefix\n\n";
 114
 115 write_file( $obo_file,  {append => 0 }, $obo_header  ) ;
 116 my $count=0;
 117 while(my $cvterm = $cvterm_rs->next() ) {
 118     my $accession = $cvterm->dbxref->accession();
 119     my $dbh = CXGN::DB::InsertDBH->new( { dbhost=>$dbhost,
 120     print STDERR "Looking at Accession $accession\n";
 121     my $cvterm_name = $cvterm->name();
 122     my $namespace = $cvterm->cv->name();
 123     my $def = $cvterm->definition();
 124     #remove quotes from definition sting
 125     $def =~ s/"//g;
 126     my $is_obsolete = $cvterm->is_obsolete();
 127     my $is_relationshiptype = $cvterm->is_relationshiptype();
 128     my $is_obsolete = $cvterm->is_obsolete();
 129
 130     my $term_details = "\n[Term]\nid: $prefix:$accession\nname: $cvterm_name\nnamespace: $namespace\n";
 131     $term_details .="def: \"$def\"\n" if $def;
 132     $term_details .="is_obsolete: true\n" if $is_obsolete;
 133     $count++;
 134     if ($is_relationshiptype) {
 135         $term_details = "
 136 [Typedef]
 137 id: $cvterm_name
 138 name: $cvterm_name
 139 "
 140     }
 141     write_file( $obo_file,  {append => 1 }, "$term_details" );
 142
 143     my $syn_rs = $cvterm->cvtermsynonyms();
 144     my $xref_rs = $cvterm->cvterm_dbxrefs();
 145     my $relationships_rs = $cvterm->cvterm_relationship_subjects();
 146
 147     while( my $synonym = $syn_rs->next() ) {
 148         my $syn_name  =  $synonym->synonym();
 149         print STDERR "synonym = $syn_name\n";
 150         my $type = $synonym->type;
 151         my $type_name ;
 152
 153         defined $type ? $type_name = $type->name : "[]";
 154
 155         #synonyms need to be quoted. Somtimes they are already quoted if loaded properly from the obo loader
 156         unless ($syn_name =~ /^\"/) { $syn_name = '"' . $syn_name . '"' ; }
 157         #xref list for synonyms
 158         unless ($syn_name =~ /$\]/) { $syn_name .= " []" ; }
 159         write_file( $obo_file,  {append => 1 }, "synonym: "  . $syn_name  . $type_name . "\n" );
 160     }
 161     while( my $xref = $xref_rs->next() ) {
 162         my $xref_acc  =  $xref->dbxref->accession();
 163         my $xref_prefix = $xref->dbxref->db->name();
 164         print STDERR "xref = $xref_prefix:$xref_acc\n";
 165         write_file( $obo_file,  {append => 1 }, "xref: "  . $xref_prefix . ":"  . $xref_acc . "\n" ) if ( $xref->is_for_definition == 0 );
 166     }
 167
 168     while( my $rel = $relationships_rs->next() ) {
 169         my $object = $rel->object();
 170         my $type = $rel->type();
 171         my $object_name = $object->name();
 172         my $object_acc = $object->dbxref->accession();
 173         my $object_acc_prefix = $object->dbxref->db->name();
 174
 175         my $type_name = $type->name();
 176
 177         my $relationship_format = "is_a:";
 178         if ($type_name ne "is_a") {
 179             $relationship_format = "relationship: $type_name";
 180         }
 181         print STDERR "$relationship_format $object_acc_prefix:$object_acc\n";
 182         write_file( $obo_file,  {append => 1 }, "$relationship_format $object_acc_prefix:$object_acc " . "! ". $object_name . "\n" );
 183     }
 184 }
 185 print STDERR "wrote $count terms to file $obo_file\n";