lib/SGN/Model/Cvterm.pm

   1
   2 =head1 NAME
   3
   4 SGN::Model::Cvterm - a simple model that provides information on cvterms
   5
   6 =head1 DESCRIPTION
   7
   8 Retrieves cv terms.
   9
  10 get_cvterm_object retrieves the term as a CXGN::Chado::Cvterm object.
  11
  12 get_cvterm_row retrieves the term as a DBIx::Class row.
  13
  14 Both function take a schema object, cvterm name and a cv name as an argument.
  15
  16 If a term is not in the database, undef is returned.
  17
  18 =head1 AUTHOR
  19
  20 Lukas Mueller
  21
  22 =cut
  23
  24 package SGN::Model::Cvterm;
  25
  26 use CXGN::Chado::Cvterm;
  27 use Data::Dumper;
  28
  29 sub get_cvterm_object {
  30     my $self = shift;
  31     my $schema = shift;
  32     my $cvterm_name = shift;
  33     my $cv_name = shift;
  34
  35     my $cv = $schema->resultset('Cv::Cv')->find( { name => $cv_name });
  36
  37     if (! $cv) {
  38         print STDERR "CV $cv_name not found. Ignoring.";
  39         return undef;
  40     }
  41     my $term = CXGN::Chado::Cvterm->new_with_term_name(
  42         $self->dbc()->dbh(),
  43         $cvterm_name,
  44         $cv->cv_id()
  45         );
  46
  47     return $term;
  48 }
  49
  50 sub get_cvterm_row {
  51     my $self = shift;
  52     my $schema = shift;
  53     my $name = shift;
  54     my $cv_name = shift;
  55
  56     my $cvterm = $schema->resultset('Cv::Cvterm')->find(
  57         {
  58             'me.name' => $name,
  59             'cv.name' => $cv_name,
  60         }, { join => 'cv' });
  61
  62     return $cvterm;
  63 }
  64
  65 sub get_cvterm_row_from_trait_name {
  66     my $self = shift;
  67     my $schema = shift;
  68     my $trait_name = shift;
  69
  70     #print STDERR $trait_name;
  71
  72     #fieldbook trait string should be "$trait_name|$dbname:$trait_accession" e.g. plant height|CO_334:0000123. substring on last occurance of |
  73     my $delim = "|";
  74     my $full_accession = substr $trait_name, rindex( $trait_name, $delim ) + length($delim);
  75     my $full_accession_length = length($full_accession) + length($delim);
  76     my $full_cvterm_name = substr($trait_name, 0, -$full_accession_length);
  77     my ( $db_name , $accession ) = split (/:/ , $full_accession);
  78
  79     #check if the trait name string does have
  80     $accession =~ s/\s+$//;
  81     $accession =~ s/^\s+//;
  82     $db_name  =~ s/\s+$//;
  83     $db_name  =~ s/^\s+//;
  84
  85     my $db_rs = $schema->resultset("General::Db")->search( { 'me.name' => $db_name });
  86     my $trait_cvterm;
  87     if ($db_rs->first()){
  88         $trait_cvterm = $schema->resultset("Cv::Cvterm")
  89         ->find({
  90              'dbxref.db_id'     => $db_rs->first()->db_id(),
  91              'dbxref.accession' => $accession
  92               },
  93               {
  94               'join' => 'dbxref'
  95               }
  96         );
  97     }
  98     return $trait_cvterm;
  99 }
 100
 101 sub get_trait_from_exact_components {
 102     my $self= shift;
 103     my $schema = shift;
 104     my $component_cvterm_ids = shift;
 105
 106     my @intersect_selects;
 107     foreach my $cvterm_id (@$component_cvterm_ids){
 108         push @intersect_selects, "SELECT object_id FROM cvterm_relationship WHERE subject_id = $cvterm_id";
 109     }
 110     push @intersect_selects, "SELECT object_id FROM cvterm_relationship GROUP BY 1 HAVING count(object_id) = ".scalar(@$component_cvterm_ids);
 111     my $intersect_sql = join ' INTERSECT ', @intersect_selects;
 112     my $h = $schema->storage->dbh->prepare($intersect_sql);
 113     $h->execute();
 114     my @trait_cvterm_ids;
 115     while(my ($trait_cvterm_id) = $h->fetchrow_array()){
 116         push @trait_cvterm_ids, $trait_cvterm_id;
 117     }
 118     if (scalar(@trait_cvterm_ids) > 1){
 119         die "More than one composed trait returned for the given set of exact componenets\n";
 120     }
 121     return $trait_cvterm_ids[0];
 122 }
 123
 124 sub get_trait_from_cvterm_id {
 125     my $schema = shift;
 126     my $cvterm_id = shift;
 127     my $format = shift; #can be 'concise' for just the name or 'extended' for name|DB:0000001
 128     if ($format eq 'concise'){
 129         $q = "SELECT name FROM cvterm WHERE cvterm_id=?;";
 130     }
 131     if ($format eq 'extended'){
 132         $q = "SELECT (((cvterm.name::text || '|'::text) || db.name::text) || ':'::text) || dbxref.accession::text FROM cvterm JOIN dbxref USING(dbxref_id) JOIN db USING(db_id) WHERE cvterm_id=?;";
 133     }
 134     my $h = $schema->storage->dbh->prepare($q);
 135     $h->execute($cvterm_id);
 136     $name = $h->fetchrow();
 137     return $name;
 138 }
 139
 140 sub _concatenate_cvterm_array {
 141     my $schema = shift;
 142     my $delimiter = shift;
 143     my $format = shift;
 144     my $first = shift;
 145     my $second = shift;
 146     #print STDERR "_concatenate_cvterm_array\n";
 147     #print STDERR Dumper $first;
 148     #print STDERR Dumper $second;
 149     my %first_hash = %$first;
 150     foreach my $f (keys %first_hash){
 151         my $ids = $first_hash{$f};
 152         foreach my $s (@$second){
 153             my @component_ids = @$ids;
 154             #print STDERR "_iterate\n";
 155             my $name = get_trait_from_cvterm_id($schema, $s, $format);
 156             my $concatenated_cvterm = $f.$delimiter.$name;
 157             push @component_ids, $s;
 158             delete $first_hash{$f};
 159             $first_hash{$concatenated_cvterm} = \@component_ids;
 160             #print STDERR Dumper \%first_hash;
 161         }
 162     }
 163     return \%first_hash;
 164 }
 165 sub get_traits_from_component_categories {
 166     my $self= shift;
 167     my $schema = shift;
 168     my $allowed_composed_cvs = shift;
 169     my $composable_cvterm_delimiter = shift;
 170     my $composable_cvterm_format = shift;
 171     my $cvterm_id_hash = shift;
 172     my %id_hash = %$cvterm_id_hash;
 173     delete @id_hash{ grep { scalar @{$id_hash{$_}} < 1 } keys %id_hash }; #remove cvtypes with no ids
 174
 175     my @ordered_id_groups;
 176     foreach my $cv_name (@$allowed_composed_cvs){
 177         push @ordered_id_groups, $id_hash{$cv_name};
 178     }
 179
 180     my $id_array_count = scalar(@ordered_id_groups);
 181     my $concatenated_cvterms;
 182     foreach (@{$ordered_id_groups[0]}){
 183         my $name = get_trait_from_cvterm_id($schema, $_, $composable_cvterm_format);
 184         $concatenated_cvterms->{$name} = [$_];
 185     }
 186     for my $n (0 .. $id_array_count-2){
 187         $concatenated_cvterms = _concatenate_cvterm_array($schema, $composable_cvterm_delimiter, $composable_cvterm_format, $concatenated_cvterms, $ordered_id_groups[$n+1]);
 188     }
 189
 190     #print STDERR "possible traits are: ".Dumper($concatenated_cvterms)."\n";
 191
 192     my @existing_traits;
 193     my @new_traits;
 194     foreach my $key (sort keys %$concatenated_cvterms){
 195         #my $existing_cvterm_name = $schema->resultset('Cv::Cvterm')->find({ name=>$key });
 196         #if ($existing_cvterm_name){
 197             #push @existing_traits, [$existing_cvterm_name->cvterm_id(), $key];
 198             #next;
 199         #}
 200         my $existing_cvterm_id = $self->get_trait_from_exact_components($schema, $concatenated_cvterms->{$key});
 201         if ($existing_cvterm_id){
 202             my $existing_name = get_trait_from_cvterm_id($schema, $existing_cvterm_id, 'extended');
 203             push @existing_traits, [$existing_cvterm_id, $existing_name];
 204             next;
 205         }
 206         push @new_traits, [ $concatenated_cvterms->{$key}, $key ];
 207     }
 208
 209     #print STDERR "existing traits are: ".Dumper(/@existing_traits)." and new traits are".Dumper(/@new_traits)."\n";
 210
 211     return {
 212       existing_traits => \@existing_traits,
 213       new_traits => \@new_traits
 214     };
 215   }
 216
 217 sub get_traits_from_components {
 218     my $self= shift;
 219     my $schema = shift;
 220     my $component_cvterm_ids = shift;
 221     my @component_cvterm_ids = @$component_cvterm_ids;
 222
 223     my $contains_cvterm_id = $self->get_cvterm_row($schema, 'contains', 'relationship')->cvterm_id();
 224
 225     my $q = "SELECT object_id FROM cvterm_relationship WHERE type_id = ? AND subject_id IN (@{[join',', ('?') x @component_cvterm_ids]}) GROUP BY 1";
 226
 227     my $h = $schema->storage->dbh->prepare($q);
 228     $h->execute($contains_cvterm_id, @component_cvterm_ids);
 229     my @trait_cvterm_ids;
 230     while(my ($trait_cvterm_id) = $h->fetchrow_array()){
 231         push @trait_cvterm_ids, $trait_cvterm_id;
 232     }
 233     return \@trait_cvterm_ids;
 234 }
 235
 236 sub get_components_from_trait {
 237     my $self= shift;
 238     my $schema = shift;
 239     my $trait_cvterm_id = shift;
 240
 241     my $contains_cvterm_id = $self->get_cvterm_row($schema, 'contains', 'relationship')->cvterm_id();
 242     my $q = "SELECT subject_id FROM cvterm_relationship WHERE object_id = $trait_cvterm_id and type_id = $contains_cvterm_id;";
 243     my $h = $schema->storage->dbh->prepare($q);
 244     $h->execute();
 245     my @component_cvterm_ids;
 246     while(my ($component_cvterm_id) = $h->fetchrow_array()){
 247         push @component_cvterm_ids, $component_cvterm_id;
 248     }
 249     return \@component_cvterm_ids;
 250 }
 251
 252 1;