Create UniProt Mapping file.

app/jobs/generate_tsvs.rb

@@ -32,8 +32,8 @@ def ensure_downloads_directory_exists
   end
 
   def public_file_path(filename)
-    desination_filename = [filename_prefix, filename].join('-')
-    File.join(downloads_dir_path, desination_filename)
+    destination_filename = [filename_prefix, filename].join('-')
+    File.join(downloads_dir_path, destination_filename)
   end
 
   def downloads_dir_path

app/jobs/generate_uniprot_mapping_tsv.rb

@@ -0,0 +1,54 @@
+class GenerateUniprotMappingTsv < GenerateTsvs
+    def perform
+      ensure_downloads_directory_exists
+      tsvs_to_generate.each do |e|
+        begin
+          tmp_file = tmp_file(e.file_name)
+          tmp_file.puts(e.headers.join("\t"))
+
+          e.objects.find_each do |object|
+
+            row = e.row_from_object(object)
+            if row[1].is_a?(Array)
+              overview_col = e.formatted_overview_col(object)
+              row[1].each do |r|
+                new_row = [row[0], r, overview_col].join("\t")
+                tmp_file.puts(new_row)
+              end
+            elsif row[1] == "N/A"
+              next
+            else
+              tmp_file.puts(row.join("\t"))
+            end
+          end
+
+          tmp_file.close
+          public_path = public_file_path(e.file_name)
+          FileUtils.cp(tmp_file.path, public_path)
+          File.chmod(0644, public_path)
+        ensure
+          tmp_file.unlink
+        end
+      end
+    end
+
+    def tsvs_to_generate
+      [UniprotMappingTsvPresenter]
+    end
+
+    def public_file_path(filename)
+      File.join(downloads_dir_path, filename)
+    end
+
+    def downloads_dir_path
+      TsvRelease.downloads_path
+    end
+
+    def release_path
+      ''
+    end
+
+    def filename_prefix
+      ''
+    end
+  end

app/presenters/uniprot_mapping_tsv_presenter.rb

@@ -0,0 +1,34 @@
+class UniprotMappingTsvPresenter
+    def self.objects
+      Gene.joins(variants: :evidence_items).where("variants.evidence_items.status!='rejected'").distinct
+    end
+
+    def self.headers
+      [
+        'civic_name',
+        'uniprot_name',
+        'gene_overview'
+      ]
+    end 
+
+    def self.row_from_object(gene)
+      [
+        gene.name,
+        Scrapers::MyGeneInfo.get_swissprot_name(gene),
+        formatted_overview_col(gene)
+      ]
+    end
+
+    def self.file_name
+      "UniprotMapping.tsv"
+    end 
+
+    def self.formatted_overview_col(gene)
+
+      eid_count = 0
+      gene.variants.each do |vs|
+        eid_count += vs.evidence_items.size 
+      end
+      "#{gene.assertions.size} clinical assertion(s) and #{eid_count} evidence item(s) for #{gene.variants.size} variant(s) curated from #{gene.sources.size} published source(s)"
+    end
+  end

config/scheduled_tasks.yml

@@ -52,6 +52,12 @@ GenerateMonthlyTsvs:
   queue: default
   class: GenerateMonthlyTsvs
 
+GenerateUniprotMappingTsv:
+  cron: '0 0 1 * *'
+  description: Update the UniprotMapping tsv
+  queue: default
+  class: GenerateUniprotMappingTsv
+
 GenerateNightlyTsvs:
   cron: '0 0 * * *'
   description: Update the nightly TSV dumps

lib/scrapers/my_gene_info.rb

@@ -36,5 +36,24 @@ def self.extract_entrez_id(data)
     def self.extract_official_name(data)
       data['hits'].first['name']
     end
+
+    private
+    def self.url_for_uniprot(gene_symbol)
+      "http://mygene.info/v2/query/?q=symbol:#{gene_symbol}&species=human&entrezonly=1&limit=1&fields=uniprot"
+    end
+
+    def self.get_swissprot_name(gene)
+      resp = Util.make_get_request(url_for_uniprot(gene.name))
+      data = JSON.parse(resp)
+      extract_swissprot_name(data)
+    end
+
+    def self.extract_swissprot_name(data)
+      if data['hits'].first != nil and data['hits'].first['uniprot'] != nil
+        data['hits'].first['uniprot']['Swiss-Prot']
+      else
+        'N/A'
+      end
+    end
   end
 end