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 +1,45 @@
+class GenerateUniprotMappingTsv < ApplicationJob
+  def perform
+    ensure_downloads_directory_exists
+    p = UniprotMappingTsvPresenter
+    begin
+      tmp_file = tmp_file(p.file_name)
+      tmp_file.puts(p.headers.join("\t"))
+
+      Gene.joins(variants: :evidence_items).distinct.find_each do |gene|  
+        rows = p.rows_from_object(gene)
+
+        rows.each do |r|
+          tmp_file.puts(r.join("\t"))
+        end
+      end
+      tmp_file.close
+      public_path = public_file_path(p.file_name)
+      FileUtils.cp(tmp_file.path, public_path)
+      File.chmod(0644, public_path)
+    ensure
+      tmp_file.unlink
+    end
+  end
+
+  private
+  def tmp_file(filename)
+    Tempfile.new(filename, File.join(Rails.root, 'tmp'))
+  end
+
+  def ensure_downloads_directory_exists
+    FileUtils.mkdir_p(downloads_dir_path)
+  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
+end

app/presenters/uniprot_mapping_tsv_presenter.rb

@@ -0,0 +1,32 @@
+class UniprotMappingTsvPresenter
+  def self.headers
+    [
+      'civic_name',
+      'uniprot_name',
+      'gene_overview'
+    ]
+  end 
+
+  def self.rows_from_object(gene)
+    swissprot_names = Array(Scrapers::MyGeneInfo.get_swissprot_name(gene))
+    formatted_overview = formatted_overview_col(gene)
+    swissprot_names.map do |n|
+      if n == "N/A"
+        nil
+      else
+        [gene.name, n, formatted_overview]
+      end
+    end.compact
+  end
+
+  def self.file_name
+    "UniprotMapping.tsv"
+  end 
+
+  def self.formatted_overview_col(gene)
+    eid_count = EvidenceItem.joins(variant: [:gene]).where("evidence_items.status != 'rejected'").where(variants: {gene: gene}).distinct.count
+    variant_count = gene.variants.joins(:evidence_items).where("evidence_items.status != 'rejected'").distinct.count
+    assertion_count = gene.assertions.where("status != 'rejected'").distinct.count
+    "#{assertion_count} clinical assertion(s) and #{eid_count} evidence item(s) for #{variant_count} variant(s) curated from #{gene.sources.size} published source(s)"
+  end
+end

config/database.yml

@@ -19,4 +19,4 @@ test:
 production:
   <<: *default
   pool: 20
-  database: civic
+  database: civic

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