bin/flag_obsolet_annot.pl

   1
   2 =head1 NAME
   3
   4 flag_obsolete_annot.pl
   5
   6 =head1 DESCRIPTION
   7
   8 Usage: perl flag_obsolete_annot.pl -H dbhost -D dbname -o outfile [-v]
   9
  10 parameters
  11
  12 =over 6
  13
  14 =item -H
  15
  16 hostname for database [required]
  17
  18 =item -D
  19
  20 database name [required]
  21
  22 =item -v
  23
  24 verbose output
  25
  26 =item -d
  27
  28 db name of the controlled vocabulary
  29
  30 =item -o
  31
  32 output file
  33
  34 =item -t
  35
  36 test mode
  37
  38 =back
  39
  40
  41 The script looks for locus and individual  ontology annotations in the database and prints out a
  42 list of annotations that require updating since the cvterms were obsoleted.
  43 Ususally running this script is a good idea after updating cvterms in the database (see sgn-tools/phenome/load_cvterms.pl)
  44 and prior to submitting an ontology association file to PO/GO.
  45
  46 =head1 AUTHOR
  47
  48 Naama Menda <nm249@cornell.edu>
  49
  50 =head1 VERSION AND DATE
  51
  52 Version 0.1, February 2008.
  53
  54 =cut
  55
  56
  57 #! /usr/bin/perl
  58 use strict;
  59
  60 use Getopt::Std;
  61
  62 use CXGN::Phenome::Locus;
  63 use CXGN::Phenome::Individual;
  64 use CXGN::Chado::Organism;
  65
  66 use CXGN::DB::InsertDBH;
  67 use CXGN::Chado::Dbxref;
  68 use CXGN::Chado::Cvterm;
  69 use CXGN::Chado::Ontology;
  70 use CXGN::Chado::Relationship;
  71
  72 our ($opt_H, $opt_D, $opt_v, $opt_o, $opt_d, $opt_t);
  73
  74 #getopts('F:d:H:o:n:vD:t');
  75 getopts('H:o:d:vD:t');
  76 my $dbhost = $opt_H;
  77 my $dbname = $opt_D;
  78
  79 if (!$dbhost && !$dbname) { die "Need -D dbname and -H hostname arguments.\n"; }
  80
  81 my $error = 0; # keep track of input errors (in command line switches).
  82 if (!$opt_D) {
  83     print STDERR "Option -D required. Must be a valid database name.\n";
  84     $error=1;
  85 }
  86
  87
  88
  89 my $file = $opt_o;
  90
  91 if (!$file) {
  92     print STDERR "A file is required as a command line argument.\n";
  93     $error=1;
  94 }
  95
  96
  97 die "Some required command lines parameters not set. Aborting.\n" if $error;
  98
  99
 100 open (OUT, ">$opt_o") ||die "can't open error file $file for writting.\n" ;
 101
 102
 103 my $dbh = CXGN::DB::InsertDBH->new( { dbhost=>$dbhost,
 104                                      dbname=>$dbname,
 105                                    } );
 106
 107
 108 print STDERR "Connected to database $dbname on host $dbhost.\n";
 109 my @locus_annot= CXGN::Phenome::Locus->get_annotations_by_db($dbh, $opt_d);
 110
 111 my $count=0;
 112 my $u_count =0;
 113 print STDERR "Reading annotations from database..\n";
 114
 115 eval {
 116     foreach my $annot(@locus_annot)  {
 117
 118         my $locus_id= $annot->get_locus_id();
 119         print STDERR "locus_id = $locus_id\n";
 120         print STDERR "." if !$opt_v;
 121         my $locus= CXGN::Phenome::Locus->new($dbh,$locus_id);
 122         my $locus_name = $locus->get_locus_name();
 123         my $dbxref=CXGN::Chado::Dbxref->new($dbh, $annot->get_dbxref_id);
 124
 125         my $cvterm = $dbxref->get_cvterm();
 126         my $cvterm_name=$cvterm->get_cvterm_name();
 127         my $accession = $dbxref->get_accession();
 128         my $is_obsolete= $cvterm->get_obsolete();
 129         my @alt_ids = $cvterm->get_alt_id();
 130         print STDERR "alt_ids = @alt_ids\n\n";
 131
 132         if ($is_obsolete) {
 133             $count++;
 134             print STDERR "Locus $locus_name (id=$locus_id) has obsolete annotation: $accession:$cvterm_name\n";
 135             print OUT "Locus $locus_name (id=$locus_id) has obsolete annotation: $accession:$cvterm_name\n";
 136
 137             if (@alt_ids) {
 138                 $u_count++;
 139                 $annot->update_annotation($alt_ids[0]);
 140                 print STDERR "*Updated annotation to $alt_ids[0]!\n";
 141                 print OUT "*Updated annotation to $alt_ids[0]\n";
 142             }else {
 143                 print STDERR "!did not find alternative cvterm for this obsolete annotation! $cvterm_name\n";
 144                 print OUT "!did not find alternative cvterm for this obsolete annotation! $cvterm_name\n";
 145             }
 146         }
 147
 148     }
 149
 150     print STDERR "Found $count obsolete annotations for SGN loci, $u_count annotations were updated.\n printed out file $file... Done.\n";
 151
 152     my @pheno_annot= CXGN::Phenome::Individual->get_annotations_by_db($dbh, $opt_d);
 153
 154     $count= 0;
 155     $u_count=0;
 156     print STDERR "Reading annotations from SGN individual database..\n";
 157     foreach my $annot(@pheno_annot) {
 158
 159         print STDERR "." if !$opt_v;
 160         my $ind_id= $annot->get_individual_id();
 161         my $ind= CXGN::Phenome::Individual->new($dbh,$ind_id);
 162         my $ind_name=$ind->get_name();
 163         my $dbxref=CXGN::Chado::Dbxref->new($dbh, $annot->get_dbxref_id());
 164         my $cvterm=$dbxref->get_cvterm();
 165         my $cvterm_name=$cvterm->get_cvterm_name();
 166         my $accession = $dbxref->get_accession();
 167         my $is_obsolete=$cvterm->get_obsolete();
 168         my @alt_ids = $cvterm->get_alt_id();
 169
 170         if ($is_obsolete) {
 171             $count++;
 172             print STDERR "Individual $ind_name (id=$ind_id) has obsolete annotation: $accession:$cvterm_name\n";
 173             print STDERR "Individual $ind_name (id=$ind_id) has obsolete annotation: $accession:$cvterm_name\n";
 174             if (@alt_ids) {
 175                 $u_count++;
 176                 $annot->update_annotation($alt_ids[0]);
 177                 print STDERR "*Updated annotation to $alt_ids[0]!\n";
 178                 print OUT "*Updated annotation to $alt_ids[0]\n";
 179             }else {
 180                 print STDERR "!did not find alternative cvterm for this obsolete annotation! $cvterm_name\n";
 181                 print OUT "!did not find alternative cvterm for this obsolete annotation! $cvterm_name\n";
 182             }
 183         }
 184     }
 185 };
 186 if ($@ || ($opt_t)) {
 187     print STDERR "Either running as trial mode (-t) or AN ERROR OCCURRED: $@\n";
 188     print OUT "Either running as trial mode (-t) or AN ERROR OCCURRED: $@\n" if $opt_o;
 189
 190     $dbh->rollback();
 191     exit(0);
 192 }
 193 else {  $dbh->commit(); }
 194
 195
 196 close OUT;
 197
 198 print STDERR "Found $count obsolete annotations for SGN individuals, $u_count annotations were updated.\n printed into out file $file... Done.\n";
 199
 200
 201