lib/CXGN/Trial/TrialLayoutDownload.pm

   1 package CXGN::Trial::TrialLayoutDownload;
   2
   3 =head1 NAME
   4
   5 CXGN::Trial::TrialLayoutDownload
   6
   7 =head1 SYNOPSIS
   8
   9 Module to format layout info for trial based on which columns user wants to see. Selected columns can be:
  10 - for plot level layout: 'plot_name','plot_id','accession_name','accession_id','plot_number','block_number','is_a_control','rep_number','range_number','row_number','col_number','seedlot_name','seed_transaction_operator','num_seed_per_plot','pedigree','location_name','trial_name','year','synonyms','tier','plot_geo_json'
  11 - for plant level layout: 'plant_name','plant_id','subplot_name','subplot_id','plot_name','plot_id','accession_name','accession_id','plot_number','block_number','is_a_control','range_number','rep_number','row_number','col_number','seedlot_name','seed_transaction_operator','num_seed_per_plot','subplot_number','plant_number','pedigree','location_name','trial_name','year','synonyms','tier','plot_geo_json'
  12 - for subplot level layout:
  13 'subplot_name','subplot_id','plot_name','plot_id','accession_name','accession_id','plot_number','block_number','is_a_control','rep_number','range_number','row_number','col_number','seedlot_name','seed_transaction_operator','num_seed_per_plot','subplot_number','pedigree','location_name','trial_name','year','synonyms','tier','plot_geo_json'
  14 - for tissue sample field trial level layout:
  15 'tissue_sample_name','tissue_sample_id','plant_name','plant_id','subplot_name','subplot_id','plot_name','plot_id','accession_name','accession_id','plot_number','block_number','is_a_control','range_number','rep_number','row_number','col_number','seedlot_name','seed_transaction_operator','num_seed_per_plot','subplot_number','plant_number','tissue_sample_number','pedigree','location_name','trial_name','year','synonyms','tier','plot_geo_json'
  16
  17 This module can also optionally include treatments into the output.
  18 This module can also optionally include accession trait performace summaries into the output.
  19
  20 This module is used from CXGN::Trial::Download::Plugin::TrialLayoutExcel, CXGN::Trial::Download::Plugin::TrialLayoutCSV, CXGN::Fieldbook::DownloadFile, CXGN::Trial->get_plots, CXGN::Trial->get_plants, CXGN::Trial->get_subplots, CXGN::Trial->get_tissue_samples
  21
  22 my $trial_layout_download = CXGN::Trial::TrialLayoutDownload->new({
  23     schema => $schema,
  24     trial_id => $trial_id,
  25     data_level => 'plots',
  26     treatment_project_ids => [1,2],
  27     selected_columns => {"plot_name"=>1,"plot_number"=>1,"block_number"=>1},
  28     selected_trait_ids => [1,2,3],
  29 });
  30 my $output = $trial_layout_download->get_layout_output();
  31
  32 Output is an ArrayRef or ArrayRefs where the first entry is the Header and subsequent entries are the layout entries.
  33
  34
  35 If you don't need treatments or phenotype summaries included you can ignore those keys like:
  36
  37 my $trial_layout_download = CXGN::Trial::TrialLayoutDownload->new({
  38     schema => $schema,
  39     trial_id => $trial_id,
  40     data_level => 'plots',
  41     selected_columns => {"plot_name"=>1,"plot_number"=>1,"block_number"=>1},
  42 });
  43 my $output = $trial_layout_download->get_layout_output();
  44
  45 Data Level can be plots, plants, subplots (for splitplot design), field_trial_tissue_samples (for seeing tissue_samples linked to plants in a field trial), plate (for seeing genotyping_layout plate tissue_samples)
  46
  47 =head1 AUTHORS
  48
  49
  50 =cut
  51
  52
  53 use Moose;
  54 use Moose::Util::TypeConstraints;
  55 use Try::Tiny;
  56 use Data::Dumper;
  57 use CXGN::Trial;
  58 use CXGN::Trial::TrialLayout;
  59 use SGN::Model::Cvterm;
  60 use CXGN::Stock;
  61 use CXGN::Stock::Accession;
  62 use JSON;
  63 use CXGN::List::Transform;
  64 use CXGN::Phenotypes::Summary;
  65 use CXGN::Phenotypes::Exact;
  66 use CXGN::Trial::TrialLayoutDownload::PlotLayout;
  67 use CXGN::Trial::TrialLayoutDownload::PlantLayout;
  68 use CXGN::Trial::TrialLayoutDownload::SubplotLayout;
  69 use CXGN::Trial::TrialLayoutDownload::TissueSampleLayout;
  70 use CXGN::Trial::TrialLayoutDownload::GenotypingPlateLayout;
  71 use CXGN::Trial::TrialLayoutDownload::SamplingTrialLayout;
  72
  73 has 'schema' => (
  74     is       => 'rw',
  75     isa      => 'DBIx::Class::Schema',
  76     required => 1,
  77 );
  78
  79 has 'trial_id'   => (
  80     isa => "Int",
  81     is => 'ro',
  82     required => 1,
  83 );
  84
  85 has 'data_level' => (
  86     is => 'ro',
  87     isa => 'Str',
  88     default => 'plots',
  89 );
  90
  91 has 'treatment_project_ids' => (
  92     isa => 'ArrayRef[Int]|Undef',
  93     is => 'rw'
  94 );
  95
  96 has 'selected_columns' => (
  97     is => 'ro',
  98     isa => 'HashRef',
  99     default => sub { {"plot_name"=>1, "plot_number"=>1} }
 100 );
 101
 102 has 'include_measured'=> (
 103     is => 'rw',
 104     isa => 'Str',
 105     default => 'false',
 106 );
 107
 108 has 'all_stats'=> (
 109     is => 'rw',
 110     isa => 'Str',
 111     default => 'true',
 112 );
 113
 114 has 'use_synonyms'=> (
 115     is => 'rw',
 116     isa => 'Str',
 117     default => 'true',
 118 );
 119
 120 has 'selected_trait_ids'=> (
 121     is => 'ro',
 122     isa => 'ArrayRef[Int]|Undef',
 123 );
 124
 125 has 'trial_stock_type'=> (
 126     is => 'rw',
 127     isa => 'Str',
 128     required => 0,
 129     default => 'accession',
 130 );
 131
 132
 133 #The attributes below are populated when get_layout_output is run, so should not be instantiatied
 134 #----------------------
 135
 136 #This is a hashref of the cached trial_layout_json that comes from CXGN::Trial::TrialLayout
 137 has 'design' => (
 138     isa => 'HashRef',
 139     is => 'rw',
 140 );
 141
 142 subtype 'Trial',
 143   as 'Ref',
 144     where { $_ =~ /CXGN::Trial/ || $_ =~ /CXGN::PhenotypingTrial/ || $_ =~  /CXGN::GenotypingTrial/ || $_ =~ /CXGN::Folder/ || $_ =~ /CXGN::CrossingTrial/ || $_ =~ /CXGN::ManagementFactor/ || $_ =~ /CXGN::SamplingTrial/},
 145   message { "The string, $_, was not a valid trial object type"};
 146
 147
 148
 149 has 'trial' => (
 150     isa => 'Trial',
 151     is => 'rw',
 152 );
 153
 154 #This treatment_info_hash contains all the info needed to make and fill the columns for the various treatments (management factors). All of these lists are in the same order.
 155 #A key called treatment_trial_list that is a arrayref of the CXGN::Trial entries that represent the treatments (management factors) in this trial
 156 #A key called treatment_trial_names_list that is an arrayref of just the treatment (management factor) names
 157 #A key called treatment_units_hash_list that is a arrayref of hashrefs where the hashrefs indicate the stocks that the treatment was applied to.
 158 has 'treatment_info_hash' => (
 159     isa => 'HashRef',
 160     is => 'rw',
 161 );
 162
 163 has 'trait_header'=> (
 164     is => 'rw',
 165     isa => 'ArrayRef[Str]|Undef',
 166 );
 167
 168 has 'exact_performance_hash' => (
 169     isa => 'HashRef',
 170     is => 'rw',
 171 );
 172
 173 has 'overall_performance_hash' => (
 174     isa => 'HashRef',
 175     is => 'rw',
 176 );
 177
 178 sub get_layout_output {
 179     my $self = shift;
 180     my $trial_id = $self->trial_id();
 181     my $schema = $self->schema();
 182     my $data_level = $self->data_level();
 183     my $include_measured = $self->include_measured();
 184     my $all_stats = $self->all_stats();
 185     my $use_synonyms = $self->use_synonyms();
 186     my %selected_cols = %{$self->selected_columns};
 187     my $treatments = $self->treatment_project_ids();
 188     my @selected_traits = $self->selected_trait_ids() ? @{$self->selected_trait_ids} : ();
 189     my %errors;
 190     my @error_messages;
 191     my $trial_stock_type = $self->trial_stock_type();
 192     print STDERR "TrialLayoutDownload for Trial id: ($trial_id) ".localtime()."\n";
 193
 194     my $trial_layout;
 195     try {
 196         my %param = ( schema => $schema, trial_id => $trial_id );
 197         if ($data_level eq 'plate'){
 198             $param{experiment_type} = 'genotyping_layout';
 199         } else {
 200             $param{experiment_type} = 'field_layout';
 201         }
 202         $trial_layout = CXGN::Trial::TrialLayout->new(\%param);
 203     };
 204     if (!$trial_layout) {
 205         #print STDERR "Trial does not have valid field design.\n";
 206         push @error_messages, "Trial does not have valid field design.";
 207         $errors{'error_messages'} = \@error_messages;
 208         return \%errors;
 209     }
 210     print STDERR "TrialLayoutDownload retrieving deisgn ".localtime."\n";
 211     my $design = $trial_layout->get_design();
 212     if (!$design){
 213         push @error_messages, "Trial does not have valid field design. Please contact us.";
 214         $errors{'error_messages'} = \@error_messages;
 215         return \%errors;
 216     }
 217
 218     if ($data_level eq 'plot_fieldMap' ) {
 219         my %hash;
 220         my @rows;
 221         my @cols;
 222         foreach my $key (keys %$design) {
 223             my $design_info = $design->{$key};
 224             my $row_num = $design_info->{row_number};
 225             my $col_num = $design_info->{col_number};
 226             my $accession = $design_info->{accession_name};
 227             $hash{$row_num}->{$col_num} = $accession;
 228             push @rows, $row_num;
 229             push @cols, $col_num;
 230         }
 231         print STDERR "TrialLayoutDownload End for Trial id: ($trial_id) ".localtime()."\n";
 232         return {output => \%hash, rows => \@rows, cols => \@cols};
 233     }
 234
 235     print STDERR "TrialLayoutDownload running stock type checks ".localtime."\n";
 236
 237     my $selected_trial = CXGN::Trial->new({bcs_schema => $schema, trial_id => $trial_id});
 238     my $has_plants = $selected_trial->has_plant_entries();
 239     my $has_subplots = $selected_trial->has_subplot_entries();
 240     my $has_tissue_samples = $selected_trial->has_tissue_sample_entries();
 241
 242     print STDERR "TrialLayoutDownload retrieving accessions ".localtime."\n";
 243
 244     my $accessions = $selected_trial->get_accessions();
 245     my @accession_ids;
 246     foreach (@$accessions){
 247         push @accession_ids, $_->{stock_id};
 248     }
 249     print STDERR "TrialLayoutDownload retrieving trait performance if requested ".localtime."\n";
 250     my $summary_values = [];
 251     if (scalar(@selected_traits)>0){
 252         my $summary = CXGN::Phenotypes::Summary->new({
 253             bcs_schema=>$schema,
 254             trait_list=>\@selected_traits,
 255             accession_list=>\@accession_ids
 256         });
 257         $summary_values = $summary->search();
 258     }
 259     my %overall_performance_hash;
 260     foreach (@$summary_values){
 261         $overall_performance_hash{$_->[0]}->{$_->[8]} = $_;
 262     }
 263
 264     my @treatment_trials;
 265     my @treatment_names;
 266     my @treatment_units_array;
 267     if ($treatments){
 268         foreach (@$treatments){
 269             my $treatment_trial = CXGN::Trial->new({bcs_schema => $schema, trial_id => $_});
 270             my $treatment_name = $treatment_trial->get_name();
 271             push @treatment_trials, $treatment_trial;
 272             push @treatment_names, $treatment_name;
 273         }
 274     }
 275     my $exact_performance_hash;
 276     if ($data_level eq 'plots') {
 277         if ($include_measured eq 'true') {
 278             print STDERR "Getting exact trait values\n";
 279             my $exact = CXGN::Phenotypes::Exact->new({
 280                 bcs_schema=>$schema,
 281                 trial_id=>$trial_id,
 282                 data_level=>'plot'
 283             });
 284             $exact_performance_hash = $exact->search();
 285             #print STDERR "Exact Performance hash is ".Dumper($exact_performance_hash)."\n";
 286         }
 287         foreach (@treatment_trials){
 288             my $treatment_units = $_ ? $_->get_observation_units_direct('plot', ['treatment_experiment']) : [];
 289             push @treatment_units_array, $treatment_units;
 290         }
 291     } elsif ($data_level eq 'plants') {
 292         if (!$has_plants){
 293             push @error_messages, "Trial does not have plants, so you should not try to download a plant level layout.";
 294             $errors{'error_messages'} = \@error_messages;
 295             return \%errors;
 296         }
 297         if ($include_measured eq 'true') {
 298             my $exact = CXGN::Phenotypes::Exact->new({
 299                 bcs_schema=>$schema,
 300                 trial_id=>$trial_id,
 301                 data_level=>'plant'
 302             });
 303             $exact_performance_hash = $exact->search();
 304         }
 305         foreach (@treatment_trials){
 306             my $treatment_units = $_ ? $_->get_observation_units_direct('plant', ['treatment_experiment']) : [];
 307             push @treatment_units_array, $treatment_units;
 308         }
 309     } elsif ($data_level eq 'subplots') {
 310         if (!$has_subplots){
 311             push @error_messages, "Trial does not have subplots, so you should not try to download a subplot level layout.";
 312             $errors{'error_messages'} = \@error_messages;
 313             return \%errors;
 314         }
 315         if ($include_measured eq 'true') {
 316             my $exact = CXGN::Phenotypes::Exact->new({
 317                 bcs_schema=>$schema,
 318                 trial_id=>$trial_id,
 319                 data_level=>'subplot'
 320             });
 321             $exact_performance_hash = $exact->search();
 322         }
 323         foreach (@treatment_trials){
 324             my $treatment_units = $_ ? $_->get_observation_units_direct('subplot', ['treatment_experiment']) : [];
 325             push @treatment_units_array, $treatment_units;
 326         }
 327     } elsif ($data_level eq 'field_trial_tissue_samples') {
 328         if (!$has_tissue_samples){
 329             push @error_messages, "Trial does not have tissue samples, so you should not try to download a tissue sample level layout.";
 330             $errors{'error_messages'} = \@error_messages;
 331             return \%errors;
 332         }
 333         if ($include_measured eq 'true') {
 334             my $exact = CXGN::Phenotypes::Exact->new({
 335                 bcs_schema=>$schema,
 336                 trial_id=>$trial_id,
 337                 data_level=>'tissue_sample'
 338             });
 339             $exact_performance_hash = $exact->search();
 340         }
 341         foreach (@treatment_trials){
 342             my $treatment_units = $_ ? $_->get_observation_units_direct('tissue_sample', ['treatment_experiment']) : [];
 343             push @treatment_units_array, $treatment_units;
 344         }
 345     } elsif ($data_level eq 'plate') {
 346         #to make the download in the header for genotyping trials more easily understood, the terms change here
 347         if (exists($selected_cols{'plot_name'})){
 348             $selected_cols{'tissue_sample_name'} = 1;
 349             delete $selected_cols{'plot_name'};
 350         }
 351         if (exists($selected_cols{'plot_number'})){
 352             $selected_cols{'well_A01'} = 1;
 353             delete $selected_cols{'plot_number'};
 354         }
 355         $selected_cols{'exported_tissue_sample_name'} = 1;
 356     }
 357
 358     print STDERR "Treatment stock hashes\n";
 359     my @treatment_stock_hashes;
 360     foreach my $u (@treatment_units_array){
 361         my %treatment_stock_hash;
 362         foreach (@$u){
 363             $treatment_stock_hash{$_->[1]}++;
 364         }
 365         push @treatment_stock_hashes, \%treatment_stock_hash;
 366     }
 367
 368     my %treatment_info_hash = (
 369         treatment_trial_list => \@treatment_trials,
 370         treatment_trial_names_list => \@treatment_names,
 371         treatment_units_hash_list => \@treatment_stock_hashes
 372     );
 373
 374     #combine sorted exact and overall trait names and if requested convert to synonyms
 375     my @exact_trait_names = sort keys %$exact_performance_hash;
 376     my @overall_trait_names = sort keys %overall_performance_hash;
 377     my @traits = (@exact_trait_names, @overall_trait_names);
 378
 379     if ($use_synonyms eq 'true') {
 380         print STDERR "Getting synonyms\n";
 381         my $t = CXGN::List::Transform->new();
 382         my $trait_id_list = $t->transform($schema, 'traits_2_trait_ids', \@traits);
 383         my @trait_ids = @{$trait_id_list->{'transform'}};
 384         my $synonym_list = $t->transform($schema, 'trait_ids_2_synonyms', $trait_id_list->{'transform'});
 385         my @missing = @{$synonym_list->{'missing'}};
 386
 387         if (scalar @missing) {
 388             #print STDERR "Traits @missing don't have synonyms. Sticking with full trait names instead\n";
 389
 390             #push @error_messages, "Traits @missing don't have synonyms. Please turn off synonym option before proceeding\n";
 391             #$errors{'error_messages'} = \@error_messages;
 392             #return \%errors;
 393         } else {
 394             @traits = @{$synonym_list->{'transform'}};
 395         }
 396     }
 397
 398
 399     my $layout_build = {
 400         schema => $schema,
 401         trial_id => $trial_id,
 402         data_level => $data_level,
 403         selected_columns => \%selected_cols,
 404         treatment_project_ids => $treatments,
 405         design => $design,
 406         trial => $selected_trial,
 407         treatment_info_hash => \%treatment_info_hash,
 408         trait_header => \@traits,
 409         exact_performance_hash => $exact_performance_hash,
 410         overall_performance_hash => \%overall_performance_hash,
 411         all_stats => $all_stats,
 412         trial_stock_type => $trial_stock_type
 413     };
 414     my $layout_output;
 415
 416     print STDERR "TrialLayoutDownload getting output object".localtime."\n";
 417
 418     if ($data_level eq 'plots' ) {
 419         $layout_output = CXGN::Trial::TrialLayoutDownload::PlotLayout->new($layout_build);
 420     }
 421     if ($data_level eq 'plants' ) {
 422         $layout_output = CXGN::Trial::TrialLayoutDownload::PlantLayout->new($layout_build);
 423     }
 424     if ($data_level eq 'subplots' ) {
 425         $layout_output = CXGN::Trial::TrialLayoutDownload::SubplotLayout->new($layout_build);
 426     }
 427     if ($data_level eq 'field_trial_tissue_samples' ) {
 428         $layout_output = CXGN::Trial::TrialLayoutDownload::TissueSampleLayout->new($layout_build);
 429     }
 430     if ($data_level eq 'plate' ) {
 431         $layout_output = CXGN::Trial::TrialLayoutDownload::GenotypingPlateLayout->new($layout_build);
 432     }
 433     if ($data_level eq 'samplingtrial' ) {
 434         $layout_output = CXGN::Trial::TrialLayoutDownload::SamplingTrialLayout->new($layout_build);
 435     }
 436
 437     print STDERR "TrialLayoutDownload retrieving output ".localtime."\n";
 438
 439     my $output = $layout_output->retrieve();
 440
 441     print STDERR "TrialLayoutDownload End for Trial id: ($trial_id) ".localtime()."\n";
 442     return {output => $output};
 443 }
 444
 445 sub _add_treatment_to_line {
 446     my $self = shift;
 447     my $treatment_stock_hashes = shift;
 448     my $line = shift;
 449     my $design_unit_name = shift;
 450     foreach (@$treatment_stock_hashes){
 451         if(exists($_->{$design_unit_name})){
 452             push @$line, 1;
 453         } else {
 454             push @$line, '';
 455         }
 456     }
 457     return $line;
 458 }
 459
 460 sub _add_overall_performance_to_line {
 461     my $self = shift;
 462     my $overall_trait_names = shift;
 463     my $line = shift;
 464     my $overall_performance_hash = shift;
 465     my $design_info = shift;
 466     my $all_stats = shift;
 467     foreach my $t (@$overall_trait_names){
 468         my $perf = $overall_performance_hash->{$t}->{$design_info->{"accession_id"}};
 469         if($perf && $all_stats eq 'true'){
 470             push @$line, "Avg: ".$perf->[3]." Min: ".$perf->[5]." Max: ".$perf->[4]." Count: ".$perf->[2]." StdDev: ".$perf->[6];
 471         } elsif($perf) {
 472             push @$line, "Avg: ".$perf->[3]." Count: ".$perf->[2];
 473         } else {
 474             push @$line, '';
 475         }
 476     }
 477     return $line;
 478 }
 479
 480 sub _get_all_pedigrees {
 481     my $self = shift;
 482     my $design = shift;
 483     my $schema = $self->schema();
 484     my %design = %{$design};
 485
 486     print STDERR "TrialLayoutDownload running get_all_pedigrees ".localtime()."\n";
 487
 488     # collect all unique accession ids for pedigree retrieval
 489     my %accession_id_hash;
 490     foreach my $key (keys %design) {
 491         $accession_id_hash{$design{$key}{'accession_id'}} = $design{$key}{'accession_name'};
 492     }
 493     my @accession_ids = keys %accession_id_hash;
 494
 495     my %pedigree_strings;
 496     if (scalar(@accession_ids)>0) {
 497         # retrieve pedigree info using batch download (fastest method), then extract pedigree strings from download rows.
 498         my $stock = CXGN::Stock->new ( schema => $schema);
 499         my $pedigree_rows = $stock->get_pedigree_rows(\@accession_ids, 'parents_only');
 500         foreach my $row (@$pedigree_rows) {
 501             my ($progeny, $female_parent, $male_parent, $cross_type) = split "\t", $row;
 502             my $string = join ('/', $female_parent ? $female_parent : 'NA', $male_parent ? $male_parent : 'NA');
 503             $pedigree_strings{$progeny} = $string;
 504         }
 505     }
 506
 507     print STDERR "TrialLayoutDownload get_all_pedigrees finished at ".localtime()."\n";
 508
 509     return \%pedigree_strings;
 510 }
 511
 512 sub _add_exact_performance_to_line {
 513     my $self = shift;
 514     my $exact_trait_names = shift;
 515     my $line = shift;
 516     my $exact_performance_hash = shift;
 517     my $observationunit_name = shift;
 518
 519     foreach my $trait (@$exact_trait_names){
 520         my $value = $exact_performance_hash->{$trait}->{$observationunit_name };
 521         if($value) {
 522             push @$line, $value
 523         } else {
 524             push @$line, '';
 525         }
 526     }
 527     return $line;
 528 }
 529
 530 1;