Extend support for secondary BioSample accessions

app/models/genome.rb

@@ -258,6 +258,10 @@ def source_attributes
     @source_attributes
   end
 
+  ##
+  # Returns registered BioSample accessions, directly from the database
+  # if the source database is +:biosample+, or through the external links
+  # if it is +:sra+
   def biosample_accessions
     case source_database.try(:to_sym)
     when :sra
@@ -267,6 +271,15 @@ def biosample_accessions
     end
   end
 
+  ##
+  # Returns all BioSample accessions, including secondary (alternative)
+  # accessions
+  def biosample_accessions_all
+    (source_hash.try(:dig, :samples) || {}).values.map do |sample|
+      sample[:biosample_accessions] || []
+    end.flatten.uniq
+  end
+
   def sra_accessions
     case source_database.to_sym
     when :sra
@@ -277,11 +290,6 @@ def sra_accessions
     end
   end
 
-  #def sequencing_experiments
-  #  @sequencing_experiments ||=
-  #    SequencingExperiment.by_biosample(biosample_accessions)
-  #end
-
   def link(acc = nil)
     acc ||= accession
     case database
@@ -410,11 +418,13 @@ def link_sequencing_experiments!
 
       # Link experiments that should be here
       self.sequencing_experiments +=
-        SequencingExperiment.where(biosample_accession: biosample_accessions)
-                            .where.not(id: sequencing_experiments.pluck(:id))
+        SequencingExperiment
+          .where(biosample_accession: biosample_accessions_all)
+          .where.not(id: sequencing_experiments.pluck(:id))
       self.sequencing_experiments +=
-        SequencingExperiment.where(biosample_accession_2: biosample_accessions)
-                            .where.not(id: sequencing_experiments.pluck(:id))
+        SequencingExperiment
+          .where(biosample_accession_2: biosample_accessions_all)
+          .where.not(id: sequencing_experiments.pluck(:id))
     end
   end
 

app/models/genome/external_resources.rb

@@ -30,12 +30,14 @@ def reload_source_json!
     when :biosample
       source_accessions.each do |acc|
         data[acc] = external_biosample_hash(acc)
-        external_biosample_to_sra(acc)
+        data[acc][:biosample_accessions].each do |acc_alt|
+          external_biosample_to_sra(acc_alt)
+        end
       end
     end
 
-    self.queued_external = nil
     self.source_json = { retrieved_at: DateTime.now, samples: data }.to_json
+    self.queued_external = nil
     save
   end
 
@@ -56,7 +58,7 @@ def external_biosample_to_sra(acc)
 
     ng = Nokogiri::XML(body)
     ng.xpath('//EXPERIMENT_SET/EXPERIMENT').map do |exp|
-      sra_acc = exp['accession'] || exp.xpath('IDENTIFIERS/PRIMARY_ID').text
+      sra_acc = exp['accession'] || exp.xpath('//IDENTIFIERS/PRIMARY_ID').text
       SequencingExperiment.find_or_create_by(sra_accession: sra_acc) do |se|
         se.external_reuse_metadata_xml = true
         se.queued_external = nil
@@ -80,17 +82,24 @@ def external_biosample_hash(acc)
   def external_biosample_hash_ebi(acc)
     uri = "https://www.ebi.ac.uk/ena/browser/api/xml/#{acc}?includeLinks=false"
     body = external_request(uri)
-    return unless body && body != '{}'
+    return unless body.present?
 
     ng = Nokogiri::XML(body)
+    sample = ng.xpath('//SAMPLE_SET/SAMPLE').first or return
     {}.tap do |hash|
       h = { api: 'EBI' }
-      h[:title] = ng.xpath('//SAMPLE_SET/SAMPLE/TITLE').text
-      h[:description] = ng.xpath('//SAMPLE_SET/SAMPLE/DESCRIPTION').text
+      h[:title] = sample.xpath('//TITLE').text
+      h[:description] = sample.xpath('//DESCRIPTION').text
       h[:attributes] = Hash[
-        ng.xpath('//SAMPLE_SET/SAMPLE/SAMPLE_ATTRIBUTES/SAMPLE_ATTRIBUTE')
+        sample.xpath('//SAMPLE_ATTRIBUTES/SAMPLE_ATTRIBUTE')
           .map { |attr| [attr.xpath('TAG').text, attr.xpath('VALUE').text] }
       ]
+      h[:biosample_accessions] = [
+        acc, sample['accession'],
+        sample.xpath('//IDENTIFIERS/PRIMARY_ID').text,
+        sample.xpath('//IDENTIFIERS/SECONDARY_ID').text,
+        sample.xpath('//EXTERNAL_ID[@namespace="BioSample"]').text
+      ].select(&:present?).uniq
       h.each { |k, v| hash[k] = h[k] if h[k].present? }
     end
   end
@@ -101,22 +110,25 @@ def external_biosample_hash_ncbi(acc)
     uri = 'https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?' \
           "db=biosample&id=#{acc}&rettype=xml&retmode=text"
     body = external_request(uri)
-    return unless body && body != '{}'
+    return unless body.present?
 
     ng = Nokogiri::XML(body)
+    sample = ng.xpath('//BioSampleSet/BioSample').first or return
     {}.tap do |hash|
       h = { api: 'NCBI' }
-      h[:title] = ng.xpath('//BioSampleSet/BioSample/Description/Title').text
-      h[:description] =
-        ng.xpath('//BioSampleSet/BioSample/Description/Comment/Paragraph').text
+      h[:title] = sample.xpath('//Description/Title').text
+      h[:description] = sample.xpath('//Description/Comment/Paragraph').text
       h[:attributes] = Hash[
-        ng.xpath('//BioSampleSet/BioSample/Attributes/Attribute')
+        sample.xpath('//Attributes/Attribute')
           .map do |attr|
             [attr['harmonized_name'] || attr['attribute_name'], attr.text]
           end
       ]
-      package = ng.xpath('//BioSampleSet/BioSample/Package').text
+      package = sample.xpath('//Package').text
       h[:attributes][:ncbi_package] = package if package.present?
+      h[:biosample_accessions] = [
+        acc, sample['accession'], sample.xpath('//Ids/Id[@db="BioSample"]').text
+      ].compact.uniq
       h.each { |k, v| hash[k] = h[k] if h[k].present? }
     end
   end

app/models/sequencing_experiment/external_resources.rb

@@ -66,15 +66,17 @@ def external_sra_to_biosample!
           '//EXPERIMENT_SET/EXPERIMENT/DESIGN/SAMPLE_DESCRIPTOR/IDENTIFIERS'
         )
       biosample_id =
-        sample_id.xpath('EXTERNAL_ID[@namespace="BioSample"]').map(&:text)
+        sample_id.xpath('//EXTERNAL_ID[@namespace="BioSample"]')
+          .first.try(:text)
       if biosample_id.present?
-        self.biosample_accession = biosample_id.first
+        self.biosample_accession = biosample_id
         self.biosample_accession_2 =
-          sample_id.xpath('PRIMARY_ID').first.try(:text)
+          sample_id.xpath('//PRIMARY_ID').first.try(:text)
       else
         self.biosample_accession =
-          sample_id.xpath('PRIMARY_ID').first.try(:text)
-        self.biosample_accession_2 = nil
+          sample_id.xpath('//PRIMARY_ID').first.try(:text)
+        self.biosample_accession_2 =
+          sample_id.xpath('//SECONDARY_ID').first.try(:text)
       end
     else
       # Unknown XML specification

app/views/genomes/_samples.html.erb

@@ -146,10 +146,12 @@
                   <% end %>
                 </div>
               <% end %>
-              <p class="lead mx-2">
-                <%= sample[:description] %>
-              </p>
+              <p class="lead mx-2"><%= sample[:description] %></p>
               <dl class="main-section">
+                <% if sample[:biosample_accessions].present? %>
+                  <dt>BioSample Accession(s)</dt>
+                  <dd><%= sample[:biosample_accessions].join(', ') %></dd>
+                <% end %>
                 <% sample[:attributes].each do |k, v| %>
                   <dt><%= k %></dt>
                   <dd><%= v %></dd>

db/schema.rb

@@ -10,7 +10,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema.define(version: 2024_11_11_163435) do
+ActiveRecord::Schema.define(version: 2024_11_12_170318) do
 
   create_table "action_text_rich_texts", force: :cascade do |t|
     t.string "name", null: false
@@ -388,6 +388,7 @@
     t.datetime "created_at", precision: 6, null: false
     t.datetime "updated_at", precision: 6, null: false
     t.datetime "retrieved_at"
+    t.string "biosample_accession_2"
     t.index ["sra_accession"], name: "index_sequencing_experiments_on_sra_accession", unique: true
   end