Include Trait Synonyms in Trial Downloads

js/source/modules/wizard-downloads.js

@@ -194,6 +194,7 @@ export function WizardDownloads(main_id,wizard){
         var level = d3.select(".wizard-download-phenotypes-level").node().value;
         var timestamp = d3.selectAll('.wizard-download-phenotypes-timestamp').property('checked')?1:0;
         var entry_numbers = d3.selectAll('.wizard-download-phenotypes-entry-numbers').property('checked')?1:0;
+        var trait_synonyms = d3.selectAll('.wizard-download-phenotypes-trait-synonyms').property('checked')?1:0;
         var intercrop = d3.selectAll('.wizard-download-phenotypes-intercrop').property('checked')?1:0;
         var outliers = d3.selectAll('.wizard-download-phenotypes-outliers').property('checked')?1:0;
         var names = JSON.stringify(d3.select(".wizard-download-phenotypes-name").node().value.split(","));
@@ -223,6 +224,7 @@ export function WizardDownloads(main_id,wizard){
             phenotype_max_value: max,
             timestamp: timestamp,
             entry_numbers: entry_numbers,
+            trait_synonyms: trait_synonyms,
             intercrop: intercrop,
             trait_contains: names,
             include_row_and_column_numbers: 1,

lib/CXGN/Phenotypes/PhenotypeMatrix.pm

@@ -229,13 +229,20 @@ has 'trait_repeat_types' => ( # returns the repeat type for every trait keyed by
     default => sub { return {} },
 );
 
+has 'include_trait_synonyms' => ( # when true, will set the column header as "SYN = TRAIT NAME|CO_xxx:yyyyyy"
+    isa => 'Bool',
+    is => 'rw',
+    default => 0
+);
+
 sub get_phenotype_matrix {
     my $self = shift;
     my $include_pedigree_parents = $self->include_pedigree_parents();
     my $include_timestamp = $self->include_timestamp;
     my $include_phenotype_primary_key = $self->include_phenotype_primary_key;
     my $include_intercrop_stocks = $self->include_intercrop_stocks;
     my $include_entry_numbers = $self->include_entry_numbers;
+    my $include_trait_synonyms = $self->include_trait_synonyms;
     my %trial_entry_numbers;
 
     $self->trait_repeat_types( $self->retrieve_trait_repeat_types() );
@@ -272,12 +279,12 @@ sub get_phenotype_matrix {
         }
     );
 
-    my ($data, $unique_traits);
+    my ($data, $unique_traits, $trait_synonyms);
     my @info;
     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');
 
     if ($self->search_type eq 'MaterializedViewTable'){
-        ($data, $unique_traits) = $phenotypes_search->search();
+        ($data, $unique_traits, $trait_synonyms) = $phenotypes_search->search();
         print STDERR "No of lines retrieved: ".scalar(@$data)."\n";
         print STDERR "Construct Pheno Matrix Start:".localtime."\n";
 
@@ -310,7 +317,10 @@ sub get_phenotype_matrix {
 
         my @sorted_traits = sort keys(%$unique_traits);
         foreach my $trait (@sorted_traits) {
-            push @line, $trait;
+            my $trait_id = $unique_traits->{$trait};
+            my $trait_syn = $trait_synonyms->{$trait_id};
+
+            push @line, defined $trait_syn && $include_trait_synonyms ? "$trait_syn = $trait" : $trait;
             if ($include_phenotype_primary_key) {
                 push @line, $trait.'_phenotype_id';
             }
@@ -474,6 +484,7 @@ sub get_phenotype_matrix {
             push @{ $obsunit_data{$d->{obsunit_stock_id}}->{$d->{trait_name} } }, $value;
         }
 
+        my %trait_synonyms;
         foreach my $d (@$data) {
             my $obsunit_id = $d->{obsunit_stock_id};
             if (!exists($seen_obsunits{$obsunit_id})) {
@@ -551,6 +562,8 @@ sub get_phenotype_matrix {
             if ($cvterm){
                 my $timestamp_value = $d->{timestamp};
                 my $value = $d->{phenotype_value};
+                my $syn = $d->{trait_synonym};
+                $trait_synonyms{$cvterm} = $syn if $syn;
                 #my $cvterm = $trait."|".$cvterm_accession;
                 # if ($include_timestamp && $timestamp_value) {
                 #     $obsunit_data{$obsunit_id}->{$cvterm} = "$value,$timestamp_value";
@@ -671,7 +684,8 @@ sub get_phenotype_matrix {
 
         my @sorted_traits = sort keys(%traits);
         foreach my $trait (@sorted_traits) {
-            push @line, $trait;
+            my $syn = $trait_synonyms{$trait};
+            push @line, defined $syn && $include_trait_synonyms ? "$syn = $trait" : $trait;
         }
         push @line, 'notes';
 

lib/CXGN/Phenotypes/Search/MaterializedViewTable.pm

@@ -379,6 +379,37 @@ sub search {
         $location_id_lookup{$r->nd_geolocation_id} = $r->description;
     }
 
+    # Get the trait ontology CVs
+    my $type_cvterm = $schema->resultset("Cv::Cvterm")->search(
+        {
+            'me.name' => { -in => ['trait_ontology', 'composed_trait_ontology'] },
+            'cv.name' => 'composable_cvtypes'
+        },
+        { join => 'cv' }
+    );
+    my @type_ids = $type_cvterm->get_column('cvterm_id')->all();
+    my $trait_cv = $schema->resultset("Cv::Cvprop")->search({ type_id => { -in => \@type_ids } });
+    my @trait_cv_ids = $trait_cv->get_column('cv_id')->all();
+
+    # Create a map of trait cvterm_ids -> synonyms
+    my %trait_synonyms;
+    my $cvtermsynonym_rs = $schema->resultset("Cv::Cvtermsynonym")->search(
+        { 
+            'cvterm.cv_id' => { -in => \@trait_cv_ids }
+        },
+        {
+            select => [ 'cvterm.cvterm_id', { min => 'synonym', -as => 'synonym' } ],
+            as => ['cvterm_id', 'synonym'],
+            join => 'cvterm',
+            group_by => [ 'cvterm.cvterm_id' ]
+        }
+    );
+    while ( my $r = $cvtermsynonym_rs->next() ) {
+        my $cvterm_id = $r->get_column('cvterm_id');
+        my $synonym = $r->get_column('synonym');
+        $trait_synonyms{$cvterm_id} = $synonym;
+    }
+
     my $h = $schema->storage->dbh()->prepare($q);
     $h->execute();
     my @result;
@@ -443,7 +474,7 @@ sub search {
 
 
 	    my $phenotype_uniquename = $o->{uniquename};
-	    $unique_traits{$trait_name}++;
+	    $unique_traits{$trait_name} = $o->{trait_id};
 	    if ($include_timestamp){
 
 		my $timestamp_value;
@@ -470,6 +501,9 @@ sub search {
 		    }
 		}
 	    }
+
+	    $o->{trait_synonym} = $trait_synonyms{$o->{trait_id}};
+
 	    push @return_observations, $o;
 	}
 
@@ -538,7 +572,7 @@ sub search {
     #print STDERR Dumper \@result;
 
     print STDERR "Search End:".localtime."\n";
-    return (\@result, \%unique_traits);
+    return (\@result, \%unique_traits, \%trait_synonyms);
 }
 
 sub _sql_from_arrayref {

lib/CXGN/Phenotypes/Search/Native.pm

@@ -372,6 +372,7 @@ sub search {
       LEFT JOIN phenotype USING(phenotype_id)
       $phenotypeprop_sql
       LEFT JOIN cvterm ON (phenotype.cvalue_id=cvterm.cvterm_id)
+      LEFT JOIN cvtermsynonym ON (cvterm.cvterm_id = cvtermsynonym.cvterm_id)
       LEFT JOIN dbxref ON (cvterm.dbxref_id = dbxref.dbxref_id)
       LEFT JOIN db USING(db_id)
       LEFT JOIN nd_experiment_project ON (nd_experiment_project.nd_experiment_id=nd_experiment_stock.nd_experiment_id)
@@ -420,6 +421,7 @@ sub search {
                         folder.description, 
                         cvterm.cvterm_id, 
                         (((cvterm.name::text || '|'::text) || db.name::text) || ':'::text) || dbxref.accession::text, 
+                        MIN(cvtermsynonym.synonym),
                         phenotype.value, 
                         phenotype.uniquename, 
                         phenotype.phenotype_id, 
@@ -433,7 +435,7 @@ sub search {
                         STRING_AGG(DISTINCT(ics.uniquename), ',')
                         ".$design_layout_select;
 
-    my $order_clause = " ORDER BY 6, 2, 29";
+    my $order_clause = " ORDER BY 6, 2, 30";
 
     my $group_by = " GROUP BY observationunit.stock_id, 
                     observationunit.uniquename, 
@@ -636,7 +638,7 @@ sub search {
 
     my $calendar_funcs = CXGN::Calendar->new({});
 
-    while (my ($observationunit_stock_id, $observationunit_uniquename, $observationunit_type_name, $germplasm_uniquename, $germplasm_stock_id, $project_project_id, $project_name, $project_description, $plot_width, $plot_length, $field_size, $field_trial_is_planned_to_be_genotyped, $field_trial_is_planned_to_cross, $breeding_program_project_id, $breeding_program_name, $breeding_program_description, $year, $design, $location_id, $planting_date, $harvest_date, $folder_id, $folder_name, $folder_description, $trait_id, $trait_name, $phenotype_value, $phenotype_uniquename, $phenotype_id, $phenotype_collect_date, $phenotype_operator, $phenotype_additional_info, $phenotype_external_references, $full_count, $notes, $intercrop_stock_id, $intercrop_stock_name, $rep_select, $block_number_select, $plot_number_select, $is_a_control_select, $row_number_select, $col_number_select, $plant_number) = $h->fetchrow_array()) {
+    while (my ($observationunit_stock_id, $observationunit_uniquename, $observationunit_type_name, $germplasm_uniquename, $germplasm_stock_id, $project_project_id, $project_name, $project_description, $plot_width, $plot_length, $field_size, $field_trial_is_planned_to_be_genotyped, $field_trial_is_planned_to_cross, $breeding_program_project_id, $breeding_program_name, $breeding_program_description, $year, $design, $location_id, $planting_date, $harvest_date, $folder_id, $folder_name, $folder_description, $trait_id, $trait_name, $trait_synonym, $phenotype_value, $phenotype_uniquename, $phenotype_id, $phenotype_collect_date, $phenotype_operator, $phenotype_additional_info, $phenotype_external_references, $full_count, $notes, $intercrop_stock_id, $intercrop_stock_name, $rep_select, $block_number_select, $plot_number_select, $is_a_control_select, $row_number_select, $col_number_select, $plant_number) = $h->fetchrow_array()) {
         my $timestamp_value;
         my $operator_value;
         if ($include_timestamp) {
@@ -733,6 +735,7 @@ sub search {
             folder_description => $folder_description,
             trait_id => $trait_id,
             trait_name => $trait_name,
+            trait_synonym => $trait_synonym,
             phenotype_value => $phenotype_value,
             phenotype_uniquename => $phenotype_uniquename,
             phenotype_id => $phenotype_id,

lib/CXGN/Trial/Download.pm

@@ -217,6 +217,7 @@ has 'include_timestamp' => (isa => 'Bool', is => 'ro', default => 0);
 has 'include_pedigree_parents' => (isa => 'Bool', is => 'ro', default => 0);
 has 'include_intercrop_stocks' => (isa => 'Bool|Undef', is => 'ro', default => 0);
 has 'include_entry_numbers' => (isa => 'Bool|Undef', is => 'ro', default => 0);
+has 'include_trait_synonyms' => (isa => 'Bool|Undef', is => 'ro', default => 0);
 has 'exclude_phenotype_outlier' => (isa => 'Bool', is => 'ro', default => 0);
 has 'has_header' => (isa => 'Bool', is => 'ro', default => 1);
 has 'trait_contains' => (isa => 'ArrayRef[Str]|Undef', is => 'rw');

lib/CXGN/Trial/Download/Plugin/TrialPhenotypeCSV.pm

@@ -94,6 +94,7 @@ sub download {
     my $include_pedigree_parents = $self->include_pedigree_parents();
     my $include_intercrop_stocks = $self->include_intercrop_stocks();
     my $include_entry_numbers = $self->include_entry_numbers();
+    my $include_trait_synonyms = $self->include_trait_synonyms();
     my $search_type = $self->search_type();
     my $phenotype_start_date = $self->start_date();
     my $phenotype_end_date = $self->end_date();
@@ -133,6 +134,7 @@ sub download {
             include_pedigree_parents=>$include_pedigree_parents,
             include_intercrop_stocks=>$include_intercrop_stocks,
             include_entry_numbers=>$include_entry_numbers,
+            include_trait_synonyms=>$include_trait_synonyms,
             phenotype_start_date => $phenotype_start_date,
             phenotype_end_date => $phenotype_end_date,
             repetitive_measurements => $repetitive_measurements,

lib/CXGN/Trial/Download/Plugin/TrialPhenotypeExcel.pm

@@ -96,6 +96,7 @@ sub download {
     my $search_type = $self->search_type();
     my $include_intercrop_stocks = $self->include_intercrop_stocks();
     my $include_entry_numbers = $self->include_entry_numbers();
+    my $include_trait_synonyms = $self->include_trait_synonyms();
     my $phenotype_start_date = $self->start_date();
     my $phenotype_end_date = $self->end_date();
     my $repetitive_measurements = $self->repetitive_measurements();
@@ -133,6 +134,7 @@ sub download {
             phenotype_max_value=>$phenotype_max_value,
             include_intercrop_stocks=>$include_intercrop_stocks,
             include_entry_numbers=>$include_entry_numbers,
+            include_trait_synonyms=>$include_trait_synonyms,
             phenotype_start_date => $phenotype_start_date,
             phenotype_end_date => $phenotype_end_date,
             repetitive_measurements => $repetitive_measurements

lib/SGN/Controller/BreedersToolbox/Download.pm

@@ -252,6 +252,7 @@ sub download_phenotypes_action : Path('/breeders/trials/phenotype/download') Arg
     my $include_pedigree_parents = $c->req->param('include_pedigree_parents');
     my $include_intercrop_stocks = $c->req->param("intercrop") && $c->req->param("intercrop") ne 'null' && $c->req->param("intercrop") ne 'undefined' ? $c->req->param("intercrop") : 0;
     my $include_entry_numbers = $c->req->param("entry_numbers") && $c->req->param("entry_numbers") ne 'null' ? $c->req->param("entry_numbers") : 0;
+    my $include_trait_synonyms = $c->req->param("trait_synonyms") && $c->req->param("trait_synonyms") ne 'null' ? $c->req->param("trait_synonyms") : 0;
     my $trait_list = $c->req->param("trait_list");
     my $trait_component_list = $c->req->param("trait_component_list");
     my $year_list = $c->req->param("year_list");
@@ -437,6 +438,7 @@ sub download_phenotypes_action : Path('/breeders/trials/phenotype/download') Arg
         include_pedigree_parents=>$include_pedigree_parents,
         include_intercrop_stocks => $include_intercrop_stocks,
         include_entry_numbers => $include_entry_numbers,
+        include_trait_synonyms => $include_trait_synonyms,
         exclude_phenotype_outlier => $exclude_phenotype_outlier,
         trait_contains => \@trait_contains_list,
         phenotype_min_value => $phenotype_min_value,
@@ -532,6 +534,7 @@ sub download_action : Path('/breeders/download_action') Args(0) {
     my $exclude_phenotype_outlier = $c->req->param("exclude_phenotype_outlier") || 0;
     my $include_intercrop_stocks = $c->req->param("intercrop") || 0;
     my $include_entry_numbers = $c->req->param("entry_numbers") || 0;
+    my $include_trait_synonyms = $c->req->param("trait_synonyms") || 0;
     my $timestamp_included = $c->req->param("timestamp") || 0;
 
     # parameters for outliers download
@@ -647,6 +650,7 @@ sub download_action : Path('/breeders/download_action') Args(0) {
             dataset_excluded_outliers=>$dataset_outliers,
             include_intercrop_stocks=>$include_intercrop_stocks,
             include_entry_numbers=>$include_entry_numbers,
+            include_trait_synonyms=>$include_trait_synonyms,
     		data_level=>$datalevel,
     	);
     	@data = $phenotypes_search->get_phenotype_matrix();