lib/CXGN/Phenotypes/PhenotypeMatrix.pm

   1 package CXGN::Phenotypes::PhenotypeMatrix;
   2
   3 =head1 NAME
   4
   5 CXGN::Phenotypes::PhenotypeMatrix - an object to handle creating the phenotype matrix. Uses SearchFactory to handle searching native database or materialized views.
   6
   7 =head1 USAGE
   8
   9 my $phenotypes_search = CXGN::Phenotypes::PhenotypeMatrix->new(
  10     bcs_schema=>$schema,
  11     search_type=>$search_type,
  12     data_level=>$data_level,
  13     trait_list=>$trait_list,
  14     trial_list=>$trial_list,
  15     program_list=>$self->program_list,
  16     folder_list=>$self->folder_list,
  17     year_list=>$year_list,
  18     location_list=>$location_list,
  19     accession_list=>$accession_list,
  20     plot_list=>$plot_list,
  21     plant_list=>$plant_list,
  22     include_timestamp=>$include_timestamp,
  23     include_pedigree_parents=>$include_pedigree_parents,
  24     exclude_phenotype_outlier=>0,
  25     trait_contains=>$trait_contains,
  26     phenotype_min_value=>$phenotype_min_value,
  27     phenotype_max_value=>$phenotype_max_value,
  28     limit=>$limit,
  29     offset=>$offset
  30 );
  31 my @data = $phenotypes_search->get_phenotype_matrix();
  32
  33 =head1 DESCRIPTION
  34
  35
  36 =head1 AUTHORS
  37
  38
  39 =cut
  40
  41 use strict;
  42 use warnings;
  43 use Moose;
  44 use Data::Dumper;
  45 use SGN::Model::Cvterm;
  46 use CXGN::Stock::StockLookup;
  47 use CXGN::Phenotypes::SearchFactory;
  48 use CXGN::BreedersToolbox::Projects;
  49
  50 has 'bcs_schema' => (
  51     isa => 'Bio::Chado::Schema',
  52     is => 'rw',
  53     required => 1,
  54 );
  55
  56 #PREFERRED MaterializedViewTable (MaterializedViewTable or Native)
  57 has 'search_type' => (
  58     isa => 'Str',
  59     is => 'rw',
  60     required => 1,
  61 );
  62
  63 #(plot, plant, or all)
  64 has 'data_level' => (
  65     isa => 'Str|Undef',
  66     is => 'ro',
  67 );
  68
  69 has 'trial_list' => (
  70     isa => 'ArrayRef[Int]|Undef',
  71     is => 'rw',
  72 );
  73
  74 has 'program_list' => (
  75     isa => 'ArrayRef[Int]|Undef',
  76     is => 'rw',
  77 );
  78
  79 has 'folder_list' => (
  80     isa => 'ArrayRef[Int]|Undef',
  81     is => 'rw',
  82 );
  83
  84 has 'trait_list' => (
  85     isa => 'ArrayRef[Int]|Undef',
  86     is => 'rw',
  87 );
  88
  89 has 'accession_list' => (
  90     isa => 'ArrayRef[Int]|Undef',
  91     is => 'rw',
  92 );
  93
  94 has 'plot_list' => (
  95     isa => 'ArrayRef[Int]|Undef',
  96     is => 'rw',
  97 );
  98
  99 has 'plant_list' => (
 100     isa => 'ArrayRef[Int]|Undef',
 101     is => 'rw',
 102 );
 103
 104 has 'subplot_list' => (
 105     isa => 'ArrayRef[Int]|Undef',
 106     is => 'rw',
 107 );
 108
 109 has 'location_list' => (
 110     isa => 'ArrayRef[Int]|Undef',
 111     is => 'rw',
 112 );
 113
 114 has 'year_list' => (
 115     isa => 'ArrayRef[Int]|Undef',
 116     is => 'rw',
 117 );
 118
 119 has 'include_pedigree_parents' => (
 120     isa => 'Bool|Undef',
 121     is => 'ro',
 122     default => 0
 123 );
 124
 125 has 'include_timestamp' => (
 126     isa => 'Bool|Undef',
 127     is => 'ro',
 128     default => 0
 129 );
 130
 131 has 'exclude_phenotype_outlier' => (
 132     isa => 'Bool',
 133     is => 'ro',
 134     default => 0
 135 );
 136
 137 has 'trait_contains' => (
 138     isa => 'ArrayRef[Str]|Undef',
 139     is => 'rw'
 140 );
 141
 142 has 'phenotype_min_value' => (
 143     isa => 'Str|Undef',
 144     is => 'rw'
 145 );
 146
 147 has 'phenotype_max_value' => (
 148     isa => 'Str|Undef',
 149     is => 'rw'
 150 );
 151
 152 has 'limit' => (
 153     isa => 'Int|Undef',
 154     is => 'rw'
 155 );
 156
 157 has 'offset' => (
 158     isa => 'Int|Undef',
 159     is => 'rw'
 160 );
 161
 162 sub get_phenotype_matrix {
 163     my $self = shift;
 164     my $include_pedigree_parents = $self->include_pedigree_parents();
 165     my $include_timestamp = $self->include_timestamp;
 166
 167     print STDERR "GET PHENOMATRIX ".$self->search_type."\n";
 168
 169     my $phenotypes_search = CXGN::Phenotypes::SearchFactory->instantiate(
 170         $self->search_type,
 171         {
 172             bcs_schema=>$self->bcs_schema,
 173             data_level=>$self->data_level,
 174             trait_list=>$self->trait_list,
 175             trial_list=>$self->trial_list,
 176             program_list=>$self->program_list,
 177             folder_list=>$self->folder_list,
 178             year_list=>$self->year_list,
 179             location_list=>$self->location_list,
 180             accession_list=>$self->accession_list,
 181             plot_list=>$self->plot_list,
 182             plant_list=>$self->plant_list,
 183             subplot_list=>$self->subplot_list,
 184             include_timestamp=>$include_timestamp,
 185             exclude_phenotype_outlier=>$self->exclude_phenotype_outlier,
 186             trait_contains=>$self->trait_contains,
 187             phenotype_min_value=>$self->phenotype_min_value,
 188             phenotype_max_value=>$self->phenotype_max_value,
 189             limit=>$self->limit,
 190             offset=>$self->offset
 191         }
 192     );
 193
 194     my ($data, $unique_traits);
 195     my @info;
 196     my @metadata_headers = ( 'studyYear', 'programDbId', 'programName', 'programDescription', 'studyDbId', 'studyName', 'studyDescription', 'studyDesign', 'plotWidth', 'plotLength', 'fieldSize', 'fieldTrialIsPlannedToBeGenotyped', 'fieldTrialIsPlannedToCross', 'plantingDate', 'harvestDate', 'locationDbId', 'locationName', 'germplasmDbId', 'germplasmName', 'germplasmSynonyms', 'observationLevel', 'observationUnitDbId', 'observationUnitName', 'replicate', 'blockNumber', 'plotNumber', 'rowNumber', 'colNumber', 'entryType', 'plantNumber');
 197
 198     if ($self->search_type eq 'MaterializedViewTable'){
 199         ($data, $unique_traits) = $phenotypes_search->search();
 200
 201         print STDERR "No of lines retrieved: ".scalar(@$data)."\n";
 202         print STDERR "Construct Pheno Matrix Start:".localtime."\n";
 203
 204         my @line = @metadata_headers;
 205         push @line, ('plantedSeedlotStockDbId', 'plantedSeedlotStockUniquename', 'plantedSeedlotCurrentCount', 'plantedSeedlotCurrentWeightGram', 'plantedSeedlotBoxName', 'plantedSeedlotTransactionCount', 'plantedSeedlotTransactionWeight', 'plantedSeedlotTransactionDescription', 'availableGermplasmSeedlotUniquenames');
 206
 207         if ($include_pedigree_parents){
 208             push @line, ('germplasmPedigreeFemaleParentName', 'germplasmPedigreeFemaleParentDbId', 'germplasmPedigreeMaleParentName', 'germplasmPedigreeMaleParentDbId');
 209         }
 210
 211         my @sorted_traits = sort keys(%$unique_traits);
 212         foreach my $trait (@sorted_traits) {
 213             push @line, $trait;
 214         }
 215         push @line, 'notes';
 216         push @info, \@line;
 217
 218         my @observationunit_ids = map { $_->{observationunit_stock_id} } @$data;
 219
 220         # retrieve treatments and add to header
 221         my $project_object = CXGN::BreedersToolbox::Projects->new( { schema => $self->bcs_schema });
 222         my ($treatment_info, $unique_treatments) = $project_object->get_treatments_by_observationunit_ids(\@observationunit_ids);
 223         my @sorted_treatments = sort keys(%$unique_treatments);
 224         foreach my $treatment (@sorted_treatments) {
 225             push @line, $treatment;
 226         }
 227
 228         push @info, \@line;
 229
 230         foreach my $obs_unit (@$data){
 231             my $entry_type = $obs_unit->{obsunit_is_a_control} ? 'check' : 'test';
 232             my $synonyms = $obs_unit->{germplasm_synonyms};
 233             my $synonym_string = $synonyms ? join ("," , @$synonyms) : '';
 234             my $available_germplasm_seedlots = $obs_unit->{available_germplasm_seedlots};
 235             my %available_germplasm_seedlots_uniquenames;
 236             foreach (@$available_germplasm_seedlots){
 237                 $available_germplasm_seedlots_uniquenames{$_->{stock_uniquename}}++;
 238             }
 239             my $available_germplasm_seedlots_uniquenames = join ' AND ', (keys %available_germplasm_seedlots_uniquenames);
 240
 241             my $trial_name = $obs_unit->{trial_name};
 242             my $trial_desc = $obs_unit->{trial_description};
 243
 244             $trial_name =~ s/\s+$//g;
 245             $trial_desc =~ s/\s+$//g;
 246
 247             my @line = ($obs_unit->{year}, $obs_unit->{breeding_program_id}, $obs_unit->{breeding_program_name}, $obs_unit->{breeding_program_description}, $obs_unit->{trial_id}, $trial_name, $trial_desc, $obs_unit->{design}, $obs_unit->{plot_width}, $obs_unit->{plot_length}, $obs_unit->{field_size}, $obs_unit->{field_trial_is_planned_to_be_genotyped}, $obs_unit->{field_trial_is_planned_to_cross}, $obs_unit->{planting_date}, $obs_unit->{harvest_date}, $obs_unit->{trial_location_id}, $obs_unit->{trial_location_name}, $obs_unit->{germplasm_stock_id}, $obs_unit->{germplasm_uniquename}, $synonym_string, $obs_unit->{observationunit_type_name}, $obs_unit->{observationunit_stock_id}, $obs_unit->{observationunit_uniquename}, $obs_unit->{obsunit_rep}, $obs_unit->{obsunit_block}, $obs_unit->{obsunit_plot_number}, $obs_unit->{obsunit_row_number}, $obs_unit->{obsunit_col_number}, $entry_type, $obs_unit->{obsunit_plant_number}, $obs_unit->{seedlot_stock_id}, $obs_unit->{seedlot_uniquename}, $obs_unit->{seedlot_current_count}, $obs_unit->{seedlot_current_weight_gram}, $obs_unit->{seedlot_box_name}, $obs_unit->{seedlot_transaction_amount}, $obs_unit->{seedlot_transaction_weight_gram}, $obs_unit->{seedlot_transaction_description}, $available_germplasm_seedlots_uniquenames);
 248
 249             if ($include_pedigree_parents) {
 250                 my $germplasm = CXGN::Stock->new({schema => $self->bcs_schema, stock_id=>$obs_unit->{germplasm_stock_id}});
 251                 my $parents = $germplasm->get_parents();
 252                 push @line, ($parents->{'mother'}, $parents->{'mother_id'}, $parents->{'father'}, $parents->{'father_id'});
 253             }
 254
 255             my $observations = $obs_unit->{observations};
 256 #            print STDERR "OBSERVATIONS =".Dumper($observations)."\n";
 257             my $include_timestamp = $self->include_timestamp;
 258             my %trait_observations;
 259             foreach (@$observations){
 260                 my $collect_date = $_->{collect_date};
 261                 my $timestamp = $_->{timestamp};
 262                 if ($include_timestamp && $timestamp) {
 263                     $trait_observations{$_->{trait_name}} = "$_->{value},$timestamp";
 264                 }
 265                 elsif ($include_timestamp && $collect_date) {
 266                     $trait_observations{$_->{trait_name}} = "$_->{value},$collect_date";
 267                 }
 268                 else {
 269                     $trait_observations{$_->{trait_name}} = $_->{value};
 270                 }
 271             }
 272             foreach my $trait (@sorted_traits) {
 273                 push @line, $trait_observations{$trait};
 274             }
 275             push @line, $obs_unit->{notes};
 276
 277
 278             # add treatment info
 279             my %unit_treatments = %{$treatment_info->{$obs_unit->{observationunit_stock_id}}};
 280             foreach my $treatment (@sorted_treatments) {
 281                 push @line, $unit_treatments{$treatment};
 282             }
 283
 284             push @info, \@line;
 285         }
 286     } else {
 287         $data = $phenotypes_search->search();
 288         #print STDERR "DOWNLOAD DATA =".Dumper($data)."\n";
 289
 290         my %obsunit_data;
 291         my %traits;
 292
 293         print STDERR "No of lines retrieved: ".scalar(@$data)."\n";
 294         print STDERR "Construct Pheno Matrix Start:".localtime."\n";
 295         my @unique_obsunit_list = ();
 296         my %seen_obsunits;
 297
 298         foreach my $d (@$data) {
 299             my $cvterm = $d->{trait_name};
 300             if ($cvterm){
 301                 my $obsunit_id = $d->{obsunit_stock_id};
 302                 if (!exists($seen_obsunits{$obsunit_id})) {
 303                     push @unique_obsunit_list, $obsunit_id;
 304                     $seen_obsunits{$obsunit_id} = 1;
 305                 }
 306
 307                 my $timestamp_value = $d->{timestamp};
 308                 my $value = $d->{phenotype_value};
 309                 #my $cvterm = $trait."|".$cvterm_accession;
 310                 if ($include_timestamp && $timestamp_value) {
 311                     $obsunit_data{$obsunit_id}->{$cvterm} = "$value,$timestamp_value";
 312                 } else {
 313                     $obsunit_data{$obsunit_id}->{$cvterm} = $value;
 314                 }
 315                 $obsunit_data{$obsunit_id}->{'notes'} = $d->{notes};
 316
 317                 my $synonyms = $d->{synonyms};
 318                 my $synonym_string = $synonyms ? join ("," , @$synonyms) : '';
 319                 my $entry_type = $d->{is_a_control} ? 'check' : 'test';
 320
 321                 my $trial_name = $d->{trial_name};
 322                 my $trial_desc = $d->{trial_description};
 323
 324                 $trial_name =~ s/\s+$//g;
 325                 $trial_desc =~ s/\s+$//g;
 326
 327                 $obsunit_data{$obsunit_id}->{metadata} = [
 328                     $d->{year},
 329                     $d->{breeding_program_id},
 330                     $d->{breeding_program_name},
 331                     $d->{breeding_program_description},
 332                     $d->{trial_id},
 333                     $trial_name,
 334                     $trial_desc,
 335                     $d->{design},
 336                     $d->{plot_width},
 337                     $d->{plot_length},
 338                     $d->{field_size},
 339                     $d->{field_trial_is_planned_to_be_genotyped},
 340                     $d->{field_trial_is_planned_to_cross},
 341                     $d->{planting_date},
 342                     $d->{harvest_date},
 343                     $d->{location_id},
 344                     $d->{location_name},
 345                     $d->{accession_stock_id},
 346                     $d->{accession_uniquename},
 347                     $synonym_string,
 348                     $d->{obsunit_type_name},
 349                     $d->{obsunit_stock_id},
 350                     $d->{obsunit_uniquename},
 351                     $d->{rep},
 352                     $d->{block},
 353                     $d->{plot_number},
 354                     $d->{row_number},
 355                     $d->{col_number},
 356                     $entry_type,
 357                     $d->{plant_number}
 358                 ];
 359                 $traits{$cvterm}++;
 360             }
 361         }
 362         #print STDERR Dumper \%plot_data;
 363         #print STDERR Dumper \%traits;
 364
 365         # retrieve treatments
 366         my $project_object = CXGN::BreedersToolbox::Projects->new( { schema => $self->bcs_schema });
 367         my ($treatment_info, $unique_treatments) = $project_object->get_treatments_by_observationunit_ids(\@unique_obsunit_list);
 368         my @sorted_treatments = sort keys(%$unique_treatments);
 369
 370         my @line = @metadata_headers;
 371
 372         my @sorted_traits = sort keys(%traits);
 373         foreach my $trait (@sorted_traits) {
 374             push @line, $trait;
 375         }
 376         push @line, 'notes';
 377         foreach my $treatment (@sorted_treatments) {
 378             push @line, $treatment;
 379         }
 380         push @info, \@line;
 381
 382         foreach my $p (@unique_obsunit_list) {
 383             my @line = @{$obsunit_data{$p}->{metadata}};
 384
 385             foreach my $trait (@sorted_traits) {
 386                 push @line, $obsunit_data{$p}->{$trait};
 387             }
 388             push @line,  $obsunit_data{$p}->{'notes'};
 389             # add treatment info
 390             my %unit_treatments = %{$treatment_info->{$p}};
 391             foreach my $treatment (@sorted_treatments) {
 392                 push @line, $unit_treatments{$treatment};
 393             }
 394             push @info, \@line;
 395         }
 396     }
 397
 398     #print STDERR Dumper \@info;
 399     print STDERR "Construct Pheno Matrix End:".localtime."\n";
 400     return @info;
 401 }
 402
 403 1;