bin/loading_scripts/LoadTomDelSGN.pm

   1 package LoadTomDelSGN;
   2
   3 use Modern::Perl;
   4 use CXGN::DB::InsertDBH;
   5 use File::Slurp;
   6 use File::Basename;
   7 use CXGN::Phenome::Locus;
   8 use Bio::Chado::Schema;
   9
  10 use Moose;
  11 with 'MooseX::Runnable';
  12 with 'MooseX::Getopt';
  13
  14 has "dbh" => (
  15     is       => 'rw',
  16     isa      => 'Ref',
  17     traits   => ['NoGetopt'],
  18     required => 0,
  19 );
  20
  21 has "dbhost" => (
  22     is            => 'rw',
  23     isa           => 'Str',
  24     required      => 1,
  25     traits        => ['Getopt'],
  26     cmd_aliases   => 'H',
  27     documentation => 'required, database host',
  28 );
  29
  30 has "dbname" => (
  31     is            => 'rw',
  32     isa           => 'Str',
  33     required      => 1,
  34     traits        => ['Getopt'],
  35     documentation => 'required, database name',
  36     cmd_aliases   => 'D',
  37 );
  38
  39 has "dirname" => (
  40     is            => 'rw',
  41     isa           => 'Str',
  42     required      => 1,
  43     traits        => ['Getopt'],
  44     documentation => 'required, input directory name with TomDel files',
  45     cmd_aliases   => 'i',
  46 );
  47
  48 has 'trial' => (
  49     is          => 'rw',
  50     isa         => 'Bool',
  51     required    => 0,
  52     default     => 0,
  53     traits      => ['Getopt'],
  54     cmd_aliases => 't',
  55     documentation =>
  56       'Test run. Rollback the transaction.',
  57 );
  58
  59 sub run {
  60     my ($self,$name) = @_;
  61
  62     my $dbh =  CXGN::DB::InsertDBH->new(
  63         {
  64             dbname =>$self->dbname,
  65             dbhost => $self->dbhost,
  66         }
  67         )->get_actual_dbh();
  68
  69     #$dbh->{AutoCommit} = 1;
  70     $self->dbh($dbh);
  71     my $schema= Bio::Chado::Schema->connect(  sub { $dbh } ,  { on_connect_do => ['SET search_path TO  public;'] }
  72         );
  73     my $dirname = $self->dirname;
  74
  75     my @files = glob "$dirname/*.pdf";
  76
  77     my $db = $schema->resultset("General::Db")->find_or_create( {
  78         name => 'TomDel',
  79         urlprefix => 'https://',
  80         url => 'solgenomics.net/ftp/TomDel-0.1/',
  81                                                                 });
  82     my $counter=0;
  83     foreach my $file (@files) {
  84       chomp $file;
  85       my $filename = basename($file);
  86       #remove the extension
  87       my ($clean_filename, $ext) = split(/\.pdf/ , $filename);
  88       my ($gene_model, $chr, $position) = split (/_/ , $clean_filename ) ;
  89             #Solyc01g005000.2_SL2.50ch01_14034.pdf
  90       #chromosome: SL2.50ch08, position: 1734498, link to genotype frequencies in tomato populations
  91             my $sgn_locusname = $gene_model;
  92             if ($gene_model =~ m/Solyc.*/) {
  93                my ($tomato_locus_name, $version)  = split (/\./ , $gene_model ) ;
  94                $sgn_locusname = $tomato_locus_name;
  95          print STDERR "Found tomato locus $sgn_locusname, $chr, $position\n";
  96       }
  97       my $locus = CXGN::Phenome::Locus->new_with_locusname($dbh, $sgn_locusname);
  98       my $locus_id = $locus->get_locus_id;
  99       if ($locus_id) {
 100         if ($locus->get_obsolete eq 't') {
 101                         print STDERR "Locus $gene_model is obsolete. Skipping. \n";
 102                         next();
 103               }
 104              } else {
 105                 print STDERR "No locus exists for ID $gene_model.\n";
 106                 next();
 107              }
 108   #add the link via dbxref
 109         #print STDERR "ADDING dbxref \n";
 110            my $dbxref = $db->find_or_create_related('dbxrefs' , {
 111             accession => $filename,
 112             description => "chromosome $chr, position: $position",
 113                                                           });
 114     #chromosome: SL2.50ch08, position: 1734498, link to genotype frequencies in tomato populations
 115      my $dbxref_id = $dbxref->dbxref_id();
 116            my $dbxref_object = CXGN::Chado::Dbxref->new($dbh, $dbxref_id);
 117            $locus->add_locus_dbxref($dbxref_object,
 118                                  undef,
 119                                  $locus->get_sp_person_id);
 120            $counter++;
 121             print STDERR "Added TomDel link dbxref_id=$dbxref_id , locus_id = $locus_id\n ";
 122     }
 123     print STDERR "Added $counter locus_dbxref links \n";
 124     if ( $self->trial) {
 125         print "Trial mode! rolling back \n";
 126         $dbh->rollback;
 127     } else {
 128         print "COMMITING\n";
 129         $dbh->commit;
 130     }
 131     return 0;
 132 }
 133
 134 return 1;