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 1.0, July 2013.
  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 Bio::Chado::Schema;
  64
  65 use CXGN::Chado::Organism;
  66
  67 use CXGN::DB::InsertDBH;
  68 use CXGN::Chado::Dbxref;
  69 use CXGN::Chado::Cvterm;
  70 use CXGN::Chado::Ontology;
  71 use CXGN::Chado::Relationship;
  72
  73 our ($opt_H, $opt_D, $opt_v, $opt_o, $opt_d, $opt_t);
  74
  75 #getopts('F:d:H:o:n:vD:t');
  76 getopts('H:o:d:vD:t');
  77 my $dbhost = $opt_H;
  78 my $dbname = $opt_D;
  79
  80 if (!$dbhost && !$dbname) { die "Need -D dbname and -H hostname arguments.\n"; }
  81
  82 my $error = 0; # keep track of input errors (in command line switches).
  83 if (!$opt_D) {
  84     print STDERR "Option -D required. Must be a valid database name.\n";
  85     $error=1;
  86 }
  87 if (!$opt_d) {
  88     print STDERR "Option -d required with a db name (PO or GO).\n";
  89     $error=1;
  90 }
  91
  92 my $file = $opt_o;
  93
  94 if (!$file) {
  95     print STDERR "A file is required as a command line argument (option -o) .\n";
  96     $error=1;
  97 }
  98
  99
 100 die "Some required command lines parameters not set. Aborting.\n" if $error;
 101
 102
 103 open (OUT, ">$opt_o") ||die "can't open error file $file for writting.\n" ;
 104
 105
 106 my $dbh = CXGN::DB::InsertDBH->new( { dbhost=>$dbhost,
 107                                      dbname=>$dbname,
 108                                    } );
 109
 110 my $schema= Bio::Chado::Schema->connect(  sub { $dbh->get_actual_dbh() } );
 111 $dbh->do('SET search_path to public');
 112
 113 print STDERR "Connected to database $dbname on host $dbhost.\n";
 114 my @locus_annot= CXGN::Phenome::Locus->get_annotations_by_db($dbh, $opt_d);
 115
 116 my $count=0;
 117 my $u_count =0;
 118 print STDERR "Reading annotations from database..\n";
 119
 120 eval {
 121     foreach my $annot(@locus_annot)  {
 122         my $locus_id= $annot->get_locus_id();
 123         print STDERR "locus_id = $locus_id\n";
 124         print STDERR "." if !$opt_v;
 125         my $locus= CXGN::Phenome::Locus->new($dbh,$locus_id);
 126         my $locus_name = $locus->get_locus_name();
 127         my $dbxref=$schema->resultset("General::Dbxref")->find( { dbxref_id => $annot->get_dbxref_id } );
 128         my $cvterm = $dbxref->search_related('cvterm')->single;
 129         my $cvterm_name=$cvterm->name;
 130         my $accession = $dbxref->accession;
 131         my $is_obsolete= $cvterm->is_obsolete;
 132         my $alt_ids_rs = $cvterm->search_related('cvterm_dbxrefs', { "db.name" => $opt_d } , { join => { "dbxref" => "db" } }  );
 133         print STDERR "found " .  $alt_ids_rs->count . " alternative ids\n\n";
 134
 135         if ($is_obsolete) {
 136             $count++;
 137             print STDERR "Locus $locus_name (id=$locus_id) has obsolete annotation: $accession:$cvterm_name\n";
 138             print OUT "Locus_id $locus_id (name = $locus_name) has obsolete annotation: $accession:$cvterm_name\n";
 139             if ($alt_ids_rs) {
 140                 my $first_alt_id = $alt_ids_rs->next;
 141                 if ($first_alt_id) {
 142                     $u_count++;
 143                     my $alt_dbxref = $first_alt_id->dbxref;
 144                     #update locus_dbxref with the alternative dbxerf_id
 145                     $annot->update_annotation($alt_dbxref->dbxref_id);
 146                     print STDERR "*Updated annotation to " .  $alt_dbxref->accession . "!\n";
 147                     print OUT "*Updated annotation to " . $alt_dbxref->accession . " \n";
 148                 }else {
 149                     print STDERR "!did not find alternative cvterm for this obsolete annotation! $cvterm_name\n";
 150                     print OUT "$locus_id : \t $opt_d : $accession: !did not find alternative cvterm for this obsolete annotation! $cvterm_name\n";
 151                 }
 152             }
 153         }
 154     }
 155
 156     print STDERR "Found $count obsolete annotations for SGN loci, $u_count annotations were updated.\n printed out file $file... Done.\n";
 157     my $pheno_rs = $schema->resultset("Stock::StockCvterm")->search(
 158         {
 159             'db.name' => $opt_d
 160         },
 161         { join => { 'cvterm' => { 'dbxref' => 'db' } },
 162         } );
 163
 164     $count= 0;
 165     $u_count=0;
 166     print STDERR "Reading annotations from SGN individual database..\n";
 167     while (my $annot = $pheno_rs->next ) {
 168         print STDERR "." if !$opt_v;
 169         my $stock_id= $annot->stock_id;
 170         my $stock= $annot->search_related('stock')->single;
 171         my $stock_name=$stock->name();
 172         my $cvterm = $annot->search_related('cvterm')->single;
 173         my $dbxref = $cvterm->search_related('dbxref')->single;
 174         my $cvterm_name=$cvterm->name();
 175         my $accession = $dbxref->accession();
 176         my $is_obsolete=$cvterm->is_obsolete();
 177         my $alt_ids_rs = $cvterm->search_related('cvterm_dbxrefs');
 178
 179         if ($is_obsolete) {
 180             $count++;
 181             print STDERR "Stock $stock_name (id=$stock_id) has obsolete annotation: $accession:$cvterm_name\n";
 182             if ($alt_ids_rs) {
 183                 $u_count++;
 184                 my $first_alt_id = $alt_ids_rs->next;
 185                 my $alt_cvterm = $first_alt_id->cvterm;
 186                 #update stock_cvterm with the alternative cvterm_id
 187                 #$annot->update( { cvterm_id => $alt_cvterm->cvterm_id } );
 188                 #print STDERR "*Updated annotation to " . $alt_cvterm->name . "!\n";
 189                 #print OUT "*Updated annotation to " . $alt_cvterm->name . " \n";
 190             }else {
 191                 print STDERR "!did not find alternative cvterm for this obsolete annotation! $cvterm_name\n";
 192                 print OUT "!did not find alternative cvterm for this obsolete annotation! $cvterm_name\n";
 193             }
 194         }
 195     }
 196 };
 197 if ($@ || ($opt_t)) {
 198     print STDERR "Either running as trial mode (-t) or AN ERROR OCCURRED: $@\n";
 199     print OUT "Either running as trial mode (-t) or AN ERROR OCCURRED: $@\n" if $opt_o;
 200     $dbh->rollback();
 201     exit(0);
 202 }
 203 else {  $dbh->commit(); }
 204
 205
 206 close OUT;
 207
 208 print STDERR "Found $count obsolete annotations for SGN individuals, $u_count annotations were updated.\n printed into out file $file... Done.\n";
 209
 210
 211