Merge pull request #1095 from twalsh-ebi/tweak/contrib_tree_config

EbiArnie · web-flow · commit 4cb39ee1635f · 2025-08-05T16:05:08.000+01:00
Facilitate config of contributing gene trees
diff --git a/modules/EnsEMBL/Draw/GlyphSet/genetree.pm b/modules/EnsEMBL/Draw/GlyphSet/genetree.pm
@@ -840,7 +840,22 @@ sub features {
       
       if ($show_exons) {
         my $ref_genetree = $tree->tree;
-        $ref_genetree = $ref_genetree->alternative_trees->{default} if $ref_genetree->ref_root_id;
+
+        if ($ref_genetree->ref_root_id) {
+
+          # If $ref_genetree is a contributing tree, we need to fetch the
+          # corresponding reference tree in order to get exon boundary data.
+          my @matching_ref_genetrees = grep {
+            !$_->ref_root_id && $_->root_id == $ref_genetree->ref_root_id
+          } values %{$ref_genetree->alternative_trees};
+
+          # Gene trees are uniquely identified by their root_id, so there can be at
+          # most one gene tree with a root_id matching $ref_genetree->ref_root_id
+          if (scalar(@matching_ref_genetrees) > 0) {
+            $ref_genetree = $matching_ref_genetrees[0];
+          }
+        }
+
         unless ($ref_genetree->{_exon_boundaries_hash}) {
           my $gtos_adaptor = $tree->adaptor->db->get_GeneTreeObjectStoreAdaptor;
           my $json_string = $gtos_adaptor->fetch_by_GeneTree_and_label($ref_genetree, 'exon_boundaries');
diff --git a/modules/EnsEMBL/Web/Command/DataExport/Output.pm b/modules/EnsEMBL/Web/Command/DataExport/Output.pm
@@ -157,7 +157,7 @@ sub process {
     $url_params->{'__clear'}        = 1;
     ## Pass parameters needed for Back button to work
     my @core_params = keys %{$hub->core_object('parameters')};
-    push @core_params, qw(export_action data_type data_action component align g1 node strain hom_id);
+    push @core_params, qw(export_action data_type data_action component align align_type clusterset_id g1 hom_id node strain);
     push @core_params, $self->config_params; 
     foreach (@core_params) {
       my @values = $component->param($_);
diff --git a/modules/EnsEMBL/Web/Component/DataExport/GeneTree.pm b/modules/EnsEMBL/Web/Component/DataExport/GeneTree.pm
@@ -41,7 +41,7 @@ sub content {
   my $view_config  = $self->view_config;
 
   my $settings = $view_config->form_fields('export');
-  $settings->{'Hidden'} = [qw(align align_type node strain)];
+  $settings->{'Hidden'} = [qw(align align_type clusterset_id node strain)];
 
   ## Add export-specific settings
   my $fields_by_format;
@@ -87,9 +87,9 @@ sub content {
 
     ## Options per format
     $fields_by_format = [{'Tree formats' => {
-                                    'Newick'    => [qw(newick_mode clusterset_id)],
-                                    'NHX'       => [qw(nhx_mode clusterset_id)],
-                                    'Text'      => [qw(scale clusterset_id)],
+                                    'Newick'    => [qw(newick_mode)],
+                                    'NHX'       => [qw(nhx_mode)],
+                                    'Text'      => [qw(scale)],
                                     'OrthoXML'  => [],
                                     'PhyloXML'  => $self->phyloxml_fields, 
                             }}];
diff --git a/modules/EnsEMBL/Web/Component/Gene/ComparaTree.pm b/modules/EnsEMBL/Web/Component/Gene/ComparaTree.pm
@@ -164,20 +164,36 @@ sub content {
   }
 
   if ($hub->type eq 'Gene') {
-    if ($tree->tree->clusterset_id ne $clusterset_id) {
-      $html .= $self->_info('Phylogenetic model selection',
-        sprintf(
-          'The phylogenetic model <i>%s</i> is not available for this tree. Showing the <i>%s</i> tree instead.', $clusterset_id, $tree->tree->clusterset_id,
-          )
-      );
+    my $tree_clusterset_id = $tree->tree->clusterset_id;
+    if ($tree_clusterset_id ne $clusterset_id) {
+      my $strain_clusterset_id = $hub->species_defs->get_config($self->hub->species, 'RELATED_TAXON');
+      # When switching between the respective consensus trees of the default and strain gene-tree view
+      # (e.g. 'default' to 'murinae'), each clusterset_id represents the consensus clusterset for its
+      # respective view, so we skip the model selection info box in such cases.
+      unless ( $strain_clusterset_id
+               &&
+               (
+                 ($clusterset_id eq 'default' && $tree_clusterset_id eq $strain_clusterset_id)
+                 ||
+                 ($clusterset_id eq $strain_clusterset_id && $tree_clusterset_id eq 'default')
+               )
+             ) {
+        $html .= $self->_info('Phylogenetic model selection',
+          sprintf(
+            'The phylogenetic model <i>%s</i> is not available for this tree. Showing the <i>%s</i> tree instead.',
+            $clusterset_id,
+            $tree_clusterset_id,
+            )
+        );
+      }
     } elsif ($tree->tree->ref_root_id) {
 
       my $text = sprintf(
           'The tree displayed here has been built with the phylogenetic model <I>%s</I>. It has then been merged with trees built with other models to give the final tree and homologies. Data shown here may be inconsistent with the rest of the comparative analyses, especially homologies.', $clusterset_id
       );
       my $rank = $tree->tree->get_tagvalue('k_score_rank');
       my $score = $tree->tree->get_tagvalue('k_score');
-      $text .= sprintf('<br/>This tree is the <b>n&deg;%d</b> closest to the final tree, with a K-distance of <b>%f</b>, as computed by <a href="http://molevol.cmima.csic.es/castresana/Ktreedist.html">Ktreedist</a>.', $rank, $score) if $rank;
+      $text .= sprintf('<br/>This tree is the <b>n&deg;%d</b> closest to the final tree, with a K-distance of <b>%f</b>, as computed by <a href="https://www.biologiaevolutiva.org/jcastresana/Ktreedist.html">Ktreedist</a>.', $rank, $score) if $rank;
       $html .= $self->_info('Phylogenetic model selection', $text);
     }
   }
diff --git a/modules/EnsEMBL/Web/Object/Gene.pm b/modules/EnsEMBL/Web/Object/Gene.pm
@@ -932,23 +932,46 @@ sub get_homologue_alignments {
 }
 
 
+sub _resolve_clusterset_ids {
+  my ($self, $strain_tree) = @_;
+
+  my $consensus_clusterset_id = $strain_tree || 'default';
+  my $clusterset_id = $self->hub->param('clusterset_id');
+  return [$consensus_clusterset_id, $clusterset_id];
+}
+
+
 sub get_GeneTree {
   my $self        = shift;
   my $compara_db  = shift || 'compara';
   my $whole_tree  = shift;
   my $strain_tree = shift;
-  my $clusterset_id = $strain_tree || $self->hub->param('clusterset_id') || 'default';
-  my $cache_key  = sprintf('_protein_tree_%s_%s_%s', $compara_db, $clusterset_id, $strain_tree);
+  my ($consensus_clusterset_id, $clusterset_id) = @{$self->_resolve_clusterset_ids($strain_tree)};
+
+  my $cache_key  = sprintf('_protein_tree_%s_%s_%s', $compara_db, $clusterset_id, $consensus_clusterset_id);
 
   if (!$self->{$cache_key}) {  
     my $args = {'stable_id' => $self->stable_id, 'cdb' => $compara_db};
     my $member  = $self->get_compara_Member($args)  || return;
     my $adaptor = $member->adaptor->db->get_adaptor('GeneTree') || return;
-    my $tree    = $adaptor->fetch_all_by_Member($member, -clusterset_id => $clusterset_id)->[0];
+
+    # We fetch the consensus tree in the first instance. This is typically
+    # the 'default' tree for the current gene-tree view, which is either
+    # the tree we want, or the reference tree of the tree we want.
+    my $tree = $adaptor->fetch_default_for_Member($member, $consensus_clusterset_id);
+
     unless ($tree) {
         $tree = $adaptor->fetch_default_for_Member($member);
     }
     return unless $tree;
+
+    # If an alternative clusterset_id has been specified that
+    # is appropriate to the current gene-tree view, it should
+    # be one of the alternative trees of the consensus tree.
+    if ($clusterset_id && exists $tree->alternative_trees->{$clusterset_id}) {
+      $tree = $tree->alternative_trees->{$clusterset_id};
+    }
+
     return $tree if $whole_tree;
     
     $tree->preload;
diff --git a/modules/EnsEMBL/Web/ViewConfig/Gene/ComparaTree.pm b/modules/EnsEMBL/Web/ViewConfig/Gene/ComparaTree.pm
@@ -155,19 +155,27 @@ sub init_form_non_cacheable {
   my $form  = $self->SUPER::init_form_non_cacheable(@_);
   my %other_clustersets;
 
+  my $page_action = $hub->referer->{'ENSEMBL_ACTION'};
+  my $consensus_clusterset_id = $hub->param('strain') || $page_action =~ /^Strain_/
+                              ? $hub->species_defs->get_config($hub->species, 'RELATED_TAXON')
+                              : 'default'
+                              ;
+
   if($hub->param('g')) {
     my $database           = $hub->database($cdb);
     my $genome_db          = $database->get_GenomeDBAdaptor->fetch_by_name_assembly($hub->species_defs->SPECIES_PRODUCTION_NAME);
 
     my $member             = $database->get_GeneMemberAdaptor->fetch_by_stable_id_GenomeDB($hub->core_params->{'g'}, $genome_db);
     my $adaptor            = $database->get_GeneTreeAdaptor;
-    my $gene_tree          = $adaptor->fetch_default_for_Member($member);
+    my $gene_tree          = $adaptor->fetch_default_for_Member($member, $consensus_clusterset_id);
+    $consensus_clusterset_id = $gene_tree->clusterset_id if $gene_tree->clusterset_id ne $consensus_clusterset_id;
     %other_clustersets     = map { $_->clusterset_id => 1 } @{$adaptor->fetch_all_linked_trees($gene_tree)};
 
-    delete $other_clustersets{'default'};
+    delete $other_clustersets{$consensus_clusterset_id};
   }
 
   if (my $dropdown = $form->get_elements_by_name('clusterset_id')->[0]) {
+    $self->_replace_default_clusterset_id_option($dropdown, $consensus_clusterset_id) if $consensus_clusterset_id ne 'default';
     $dropdown->add_option({ 'value' => $_, 'caption' => $_ }) for sort keys %other_clustersets;
   }
 
@@ -179,4 +187,13 @@ sub _groups {
   return (@{ $_[0]->species_defs->TAXON_ORDER });
 }
 
+sub _replace_default_clusterset_id_option {
+  my ($self, $cset_id_dropdown, $clusterset_id) = @_;
+
+  my $default_option = $cset_id_dropdown->get_elements_by_attribute({'value' => 'default'})->[0];
+  my $cset_id_option = $cset_id_dropdown->dom->create_element('option', {'value' => $clusterset_id, 'inner_HTML' => 'Final (merged) tree'});
+  $default_option->after($cset_id_option);
+  $default_option->remove();
+}
+
 1;
