Facilitate CACTUS_DB alignment views in all divisions

modules/EnsEMBL/Web/Component/Compara_Alignments.pm

@@ -68,6 +68,15 @@ sub content {
   my ($align, $target_species, $target_slice_name_range) = split '--', $align_param;
   my $target_slice = $object->get_target_slice;
 
+  ## Check for wrong component in per-component polyploid alignments.
+  my $warning_box = $self->check_for_wrong_genome_component({
+    'cdb'   => $cdb,
+    'align' => $align,
+    'slice' => $slice,
+  });
+  return $warning_box if $warning_box;
+  ##
+
   my ($alert_box, $error) = $self->check_for_align_problems({
                     'align' => $align,
                     'species' => $hub->species_defs->SPECIES_PRODUCTION_NAME,
@@ -339,6 +348,9 @@ sub check_for_missing_species {
   return $warnings ? ({'severity' => 'info', 'title' => $title, 'message' => $warnings}) : ();
 }
 
+# Stub for use in divisions with polyploid genomes (e.g. Plants)
+sub check_for_wrong_genome_component {}
+
 sub show_warnings {
   my ($self, $messages) = @_;
   return '' unless defined $messages;

modules/EnsEMBL/Web/Component/Location/Compara_AlignSliceBottom.pm

@@ -77,6 +77,10 @@ sub content {
   my %aligned_species = map { $_->{'name'} => 1 } @$slices;
   my $i               = 1;
   my (@images, $html);
+
+  if ($align_details->{'type'} eq 'CACTUS_DB') {
+    $html .= $self->show_scale_dependent_track_info_box($align_details);
+  }
 
   my ($caption_height,$caption_img_offset) = (0,-24);
   my $lookup = $species_defs->prodnames_to_urls_lookup;
@@ -141,6 +145,55 @@ sub content {
   return $html;
 }
 
+sub show_scale_dependent_track_info_box {
+  my ($self, $align_details) = @_;
+  my $species_defs = $self->hub->species_defs;
+
+  my $html;
+  if (exists $align_details->{'as_track_threshold_data'}) {
+    my $r = $self->param('r');
+
+    my $location_length;
+    if ($r =~ /^[\w\.\-]+:(\d+)\-(\d+)$/) {  # region pattern from MetaKeyFormat datacheck
+      $location_length = abs($2 - $1) + 1;
+    } else {
+      $location_length = 1;  # This should never happen, but if it does, we revert to default behaviour.
+    }
+
+    my $as_track_thresholds = $align_details->{'as_track_threshold_data'};
+    if (exists $as_track_thresholds->{'transcript'} && $location_length >= $as_track_thresholds->{'transcript'}) {
+
+      my @range_vis_info_parts = (
+        sprintf(
+          'gene tracks hidden in regions larger than %d kb',
+          $self->thousandify($as_track_thresholds->{'transcript'}),
+        )
+      );
+
+      if (exists $as_track_thresholds->{'sequence'}) {
+        push(
+          @range_vis_info_parts,
+          sprintf(
+            'contig tracks hidden in regions larger than %d kb',
+            $self->thousandify($as_track_thresholds->{'sequence'}),
+          )
+        );
+      }
+
+      my $help_id = { $species_defs->multiX('ENSEMBL_HELP') }->{'Location/Compara_Alignments/Image'};
+      $html .= $self->_info('Scale-dependent alignment track configuration',
+          '<p>Some tracks in this Cactus image alignment are disabled by default at larger scales, with '
+          . join(', and ', @range_vis_info_parts)
+          . '. Tracks hidden in this way can be revealed by zooming in, or by enabling them directly'
+          . ' via "<strong>Configure this page</strong>" or "<strong>Add/remove tracks</strong>".'
+          . sprintf(' For more information, see the <a href="/Help/View?id=%d" class="popup">Alignments (image) help page</a>.', $help_id)
+      );
+    }
+  }
+
+  return $html;
+}
+
 sub export_options { return {'action' => 'Alignments', 'caption' => 'Download alignment'}; }
 
 sub get_export_data {

modules/EnsEMBL/Web/ConfigPacker.pm

@@ -1250,6 +1250,7 @@ sub _summarise_compara_db {
   my $constrained_elements = {};
   my %valid_species = map { $_ => 1 } keys %{$self->full_tree};
 
+  my $cactus_db_found = 0;
   foreach my $row (@$res_aref) { 
     my ($class, $type, $species, $name, $id, $species_set_id) = ($row->[0], uc $row->[1], $row->[2], $row->[3], $row->[4], $row->[5]);
     my $key = 'ALIGNMENTS';
@@ -1262,6 +1263,8 @@ sub _summarise_compara_db {
       $constrained_elements->{$species_set_id} = $id;
     } elsif ($type !~ /EPO_LOW_COVERAGE/ && ($class =~ /tree_alignment/ || $type  =~ /EPO/)) {
       $self->db_tree->{$db_name}{$key}{$id}{'species'}{'ancestral_sequences'} = 1 unless exists $self->db_tree->{$db_name}{$key}{$id};
+    } elsif ($type eq 'CACTUS_DB') {
+      $cactus_db_found = 1;
     }
 
     if ($intra_species{$species_set_id}) {
@@ -1275,6 +1278,19 @@ sub _summarise_compara_db {
     $self->db_tree->{$db_name}{$key}{$id}{'species_set_id'}    = $species_set_id;
     $self->db_tree->{$db_name}{$key}{$id}{'species'}{$species} = 1;
   }
+
+  if ($cactus_db_found) {
+    $res_aref = $dbh->selectall_arrayref('
+      select method_link_species_set_id, value
+        from method_link_species_set_tag
+      where tag = "align_slice_track_threshold_data"
+    ');
+
+    foreach my $row (@$res_aref) {
+      my ($alignment_id, $as_track_threshold_json) = ($row->[0], $row->[1]);
+      $self->db_tree->{$db_name}{'ALIGNMENTS'}{$alignment_id}{'as_track_threshold_data'} = from_json($as_track_threshold_json);
+    }
+  }
 
   foreach my $species_set_id (keys %$constrained_elements) {
     my $constr_elem_id = $constrained_elements->{$species_set_id};

modules/EnsEMBL/Web/Document/HTML/Compara.pm

@@ -333,15 +333,26 @@ sub get_species_info {
     }
   }
 
-  ## Lookup table from species name to genome_db
-  my $genome_db_name_hash = {};
+  ## Lookup table from species name to genome_db stats
+  my $genome_db_stats_hash = {};
   if ($mlss) {
+    my $genome_db_id_2_node_hash = $mlss && $mlss->species_tree && $mlss->species_tree->get_genome_db_id_2_node_hash;
     foreach my $genome_db (@{$mlss->species_set->genome_dbs}) {
+      ## Set coverage stats from species-tree node tag or MLSS tag, as available
+      my $id = $genome_db->dbID;
       my $species_tree_name = $genome_db->name;
-      $genome_db_name_hash->{$species_tree_name} = $genome_db;
+      $genome_db_stats_hash->{$species_tree_name}{'assembly'} = $genome_db->assembly;
+      my @stats = qw(genome_coverage genome_length coding_exon_coverage coding_exon_length);
+      foreach (@stats) {
+        if ($genome_db_id_2_node_hash && exists $genome_db_id_2_node_hash->{$id}
+              && defined $genome_db_id_2_node_hash->{$id}->get_value_for_tag($_)) {
+          $genome_db_stats_hash->{$species_tree_name}{$_} = $genome_db_id_2_node_hash->{$id}->get_value_for_tag($_);
+        } elsif (defined $mlss->get_value_for_tag($_.'_'.$id)) {
+          $genome_db_stats_hash->{$species_tree_name}{$_} = $mlss->get_value_for_tag($_.'_'.$id);
+        }
+      }
     }
   }
-  my $genome_db_id_2_node_hash = $mlss && $mlss->species_tree && $mlss->species_tree->get_genome_db_id_2_node_hash;
 
   ## Now munge information for selected species
   foreach my $sp (@$species_order) {
@@ -357,17 +368,9 @@ sub get_species_info {
 
     if ($mlss) {
       my $prod_name = $hub->species_defs->get_config($sp, 'SPECIES_PRODUCTION_NAME');
-      my $gdb = $genome_db_name_hash->{$prod_name};
-      $info->{$sp}{'assembly'} = $gdb->assembly;
-      ## Add coverage stats
-      my $id = $gdb->dbID;
-      my @stats = qw(genome_coverage genome_length coding_exon_coverage coding_exon_length);
-      foreach (@stats) {
-        if ($genome_db_id_2_node_hash && exists $genome_db_id_2_node_hash->{$id} && defined $genome_db_id_2_node_hash->{$id}->get_value_for_tag($_)) {
-            $info->{$sp}{$_} = $genome_db_id_2_node_hash->{$id}->get_value_for_tag($_);
-        } else {
-            $info->{$sp}{$_} = $mlss->get_value_for_tag($_.'_'.$id);
-        }
+      ## Add assembly name and coverage stats
+      while ( my ($key, $value) = each(%{$genome_db_stats_hash->{$prod_name}}) ) {
+        $info->{$sp}{$key} = $value;
       }
     }
   }

modules/EnsEMBL/Web/ImageConfig/alignsliceviewbottom.pm

@@ -101,19 +101,6 @@ sub init_cacheable {
     [ 'variation_legend' ],
     { accumulate => 'yes' }
   );
-
-  my $align_params = $self->hub->referer->{'params'}{'align'}[0];
-  my ($align) = split '--', $align_params;
-  my $align_type = $self->species_defs->multi_hash->{'DATABASE_COMPARA'}{'ALIGNMENTS'}{$align}{'type'};
-  if ($align_type eq 'CACTUS_DB') {
-    my $node = $self->get_node('transcript');
-    my @transcript_tracks = grep { $_->get_data('node_type') eq 'track' } @{$node->get_all_nodes};
-    foreach my $transcript_track (@transcript_tracks) {
-      if ($species ne $self->hub->referer->{'ENSEMBL_SPECIES'}) {
-        $transcript_track->set_data('display', 'off');
-      }
-    }
-  }
 }
 
 sub species_list {

modules/EnsEMBL/Web/ImageConfigExtension/Tracks.pm

@@ -469,7 +469,7 @@ sub add_genes {
   }
 
   # Adding gencode basic track, this has been moved from each image config to this generic one
-  if (my $gencode_version = $self->species_defs->GENCODE_VERSION || "") {
+  if (my $gencode_version = $self->species_defs->get_config($species, 'GENCODE_VERSION') || "") {
     $self->add_track('transcript', 'gencode_basic', "Basic Gene Annotations from $gencode_version", '_gencode_basic', {
       labelcaption  => "Genes (Basic set from $gencode_version)",
       display       => 'off',
@@ -492,7 +492,7 @@ sub add_genes {
   }
 
   # Adding gencode primary track
-  if (my $gencode_version = $self->species_defs->GENCODE_VERSION || "") {
+  if (my $gencode_version = $self->species_defs->get_config($species, 'GENCODE_VERSION') || "") {
     $self->add_track('transcript', 'gencode_primary', "Primary Gene Annotations from $gencode_version", '_gencode_primary', {
       labelcaption  => "Genes (Primary set from $gencode_version)",
       display       => 'transcript_label',
@@ -513,9 +513,9 @@ sub add_genes {
       ],   
     });  
   }
-  
+
   # Adding MANE tracks (Only for Humans)
-  if($self->hub->species_defs->SEPARATE_MANE_TRACKS){
+  if($self->species_defs->get_config($species, 'SEPARATE_MANE_TRACKS')){
 
     # Adding MANE Select track
     $self->add_track('transcript', 'mane_select', "MANE Select Transcripts", '_mane_select', {
@@ -557,7 +557,7 @@ sub add_genes {
   # Need to add the gene menu track here
   $self->add_track('information', 'gene_legend', 'Gene Legend', 'gene_legend', { strand => 'r' }) if $flag;
 
-  if($self->species_defs->GENCODE_VERSION) {
+  if($self->species_defs->get_config($species, 'GENCODE_VERSION')) {
     # Disable comprehensive geneset track and enable primary gencode ones
     $self->modify_configs(['transcript_core_ensembl'],{ 'display' => 'off' });
     $self->modify_configs(['gencode_primary'], { 'display' => 'transcript_label' });
@@ -566,6 +566,47 @@ sub add_genes {
     $self->modify_configs(['transcript_core_ensembl'],{ description => 'The <a class="popup" href="/Help/Glossary?id=487">GENCODE Comprehensive</a> set is the gene set for human and mouse' });
   }
 
+  # If this is an alignslice track in a large-scale CACTUS_DB alignment
+  # view, disable selected tracks in order to reduce load times.
+  if ($self->type eq 'alignsliceviewbottom') {
+
+    my $align_id = exists $self->hub->referer->{'params'}{'align'}
+                 ? $self->hub->referer->{'params'}{'align'}[0]
+                 : $self->hub->get_alignment_id
+                 ;
+
+    if ($align_id) {
+      my $align_details = $self->species_defs->multi_hash->{'DATABASE_COMPARA'}->{'ALIGNMENTS'}->{$align_id};
+      if ($align_details->{'type'} eq 'CACTUS_DB' && exists $align_details->{'as_track_threshold_data'}) {
+        my $location_param = $self->hub->referer->{'params'}{'r'}[0];
+
+        my $location_length;
+        if ($location_param =~ /^[\w\.\-]+:(\d+)\-(\d+)$/) {  # region pattern from MetaKeyFormat datacheck
+          $location_length = abs($2 - $1) + 1;
+        } else {
+          $location_length = 1;  # This should never happen, but if it does, we revert to default behaviour.
+        }
+
+        my $as_track_thresholds = $align_details->{'as_track_threshold_data'};
+        if (exists $as_track_thresholds->{'transcript'} && $location_length >= $as_track_thresholds->{'transcript'}) {
+
+          my $transcript_node_ids = $self->species_defs->get_config($species, 'GENCODE_VERSION')
+                                  ? ['gencode_primary', 'transcript']
+                                  : ['transcript']
+                                  ;
+
+          # At large scales, disable transcript tracks.
+          $self->modify_configs($transcript_node_ids, { 'display' => 'off' });
+
+          if (exists $as_track_thresholds->{'sequence'} && $location_length >= $as_track_thresholds->{'sequence'}) {
+            # At larger scales still, disable sequence tracks.
+            $self->modify_configs(['sequence'], { 'display' => 'off' });
+          }
+        }
+      }
+    }
+  }
+
 }
 
 sub add_trans_associated {