bin/dump_sgn_loci.pl

   1
   2 =head1 NAME
   3
   4 dump_sgn_loci.pl
   5
   6 =head1 DESCRIPTION
   7
   8 Usage: perl dump_sgn_loci.pl -H dbhost -D dbname -o outfile [-n common_name] [-v]
   9
  10 parameters
  11
  12 =over 5
  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 -n
  27
  28 optional- common_name. Limit results to one organism (e.g. tomato)
  29
  30 =item -o
  31
  32 output file
  33
  34
  35 =back
  36
  37
  38 The script dumps sgn loci into a tab delimited file
  39 common_name  locus_id   locus-name   locus_symbol  list of genbank sequence annotations | list of SGN-unigene ids (genbank|SGN-U)
  40
  41
  42 =head1 AUTHOR
  43
  44 Naama Menda <nm249@cornell.edu>
  45
  46 =head1 VERSION AND DATE
  47
  48 Version 0.2, March 2012.
  49
  50 =cut
  51
  52
  53 #! /usr/bin/perl
  54 use strict;
  55
  56 use Getopt::Std;
  57
  58 use CXGN::Phenome::Locus;
  59 use CXGN::Chado::Organism;
  60
  61 use CXGN::DB::InsertDBH;
  62 use CXGN::Chado::Dbxref;
  63
  64 use File::Slurp;
  65 use Bio::Chado::Schema;
  66
  67 our ($opt_H, $opt_D, $opt_v, $opt_o, $opt_n);
  68
  69 getopts('D:H:n:o:f');
  70 my $dbhost = $opt_H;
  71 my $dbname = $opt_D;
  72
  73 if (!$dbhost && !$dbname) { die "Need -D dbname and -H hostname arguments.\n"; }
  74
  75 my $error = 0; # keep track of input errors (in command line switches).
  76 if (!$opt_D) {
  77     print STDERR "Option -D required. Must be a valid database name.\n";
  78     $error=1;
  79 }
  80
  81 print STDERR "$opt_D, $opt_H, $opt_n, $opt_o\n";
  82 my $file = $opt_o;
  83
  84 if (!$file) {
  85     print STDERR "A file is required as a command line argument.\n";
  86     $error=1;
  87 }
  88
  89 die "Some required command lines parameters not set. Aborting.\n" if $error;
  90
  91 open (OUT, ">$opt_o") ||die "can't open error file $file for writting.\n" ;
  92
  93
  94 my $dbh = CXGN::DB::InsertDBH->new( { dbhost=>$dbhost,
  95                                      dbname=>$dbname,
  96                                    } );
  97
  98
  99 print STDERR "Connected to database $dbname on host $dbhost.\n";
 100 my $query = "SELECT locus_id FROM phenome.locus";
 101 $query .= " WHERE common_name_id = (SELECT common_name_id FROM sgn.common_name where common_name ilike ?) " if $opt_n;
 102 $query .= " ORDER BY locus_id";
 103 my $common_name = $opt_n || undef;
 104 my $count=0;
 105 my $sth=$dbh->prepare($query);
 106 $sth->execute($common_name);
 107 print OUT "common_name\tlocus_id\tlocus_name\tlocus_symbol\tgene_model\tSGN-unigenes\tGenBank accessions\t\tGO_annotations\tPO_annotations\tSP_annotations\n";
 108 while (my ($locus_id) = $sth->fetchrow_array()) {
 109     my $count++;
 110     my $locus=CXGN::Phenome::Locus->new($dbh, $locus_id) ;
 111     my $symbol=$locus->get_locus_symbol();
 112     my $name= $locus->get_locus_name();
 113     my $common_name = $locus->get_common_name();
 114     my @u_objects= $locus->get_unigenes( {current=>1} ); #unigene ids
 115     my @unigenes = map {'SGN-U' . $_->get_unigene_id()  } @u_objects;
 116     my $unigene_string= join '|', @unigenes;
 117     my %dbxrefs= $locus->get_dbxref_lists(); #dbxref objects
 118
 119     my @gb_accs= map {$_->[0]->get_feature->get_uniquename()  }  @{ $dbxrefs{'DB:GenBank_GI'} };
 120     my $gb_string = join '|', @gb_accs;
 121     my $genome_locus = $locus->get_genome_locus;
 122     my @go =  map { 'GO:' . $_->[0]->get_accession  }  @{ $dbxrefs{'GO'} };
 123     my $go_string = join '|' , @go;
 124     my @po =  map {'PO:' . $_->[0]->get_accession  }  @{ $dbxrefs{'PO'} };
 125     my $po_string = join '|' , @po;
 126     my @sp =  map {'SP:' . $_->[0]->get_accession  }  @{ $dbxrefs{'SP'} };
 127     my $sp_string = join '|' , @sp;
 128
 129     print OUT "$common_name\t$locus_id\t$name\t$symbol\t$genome_locus\t$unigene_string\t$gb_string\t$go_string\t$po_string\t$sp_string \n";
 130     print STDERR "$common_name \t $locus_id \t $name \t $symbol \t $genome_locus\t $unigene_string \t $gb_string \n";
 131
 132 }
 133
 134
 135 close OUT;
 136
 137 print STDERR "Found $count loci.\n printed into out file $file... Done.\n";
 138
 139
 140