rbbt-sources 3.0.36 → 3.0.37

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 6dc886fa1c1e4052a8c5dee351f46e2c2f536e39
-  data.tar.gz: 7a9722d19761ed23467f5f812ebb18787aeee534
+  metadata.gz: 27d9b524e74122ec28214154aae2646350075b34
+  data.tar.gz: 5a41416cadec47a6a65b5dc0fd15f836d8aaf27c
 SHA512:
-  metadata.gz: 7da4beb09820957334fed2d9bfaa8af9da5b206fd3f4916ca357ab48224eb266c6711b72d920eb83cdb1f9ef2080db9665515e0bfa86ac7c7c557dd427c5f17f
-  data.tar.gz: 3576bde597dd51ed3af48c20f8bcf46c57a3ae3e240bf38cc1e5e6b4173c0d9557c1375646b1de9534f2a76dac3a32fd692c84f1404c60a961ec00a372c5b98d
+  metadata.gz: f7479d80768eecbd609d4bb15907a36eb142d693ce185122fe7c4fe90d8b66cafa10aa0454dabc22db0a6ef4f73500d60752ef6f317b81c495929bfb7756bd2b
+  data.tar.gz: c9227fd124bad413ca05292702b3adcd9fe10bacbbb242022621c64c9fa63cf14ff218158dc93a48b6805e27872fb06e22f24f9cc00555896a12ef7566999037

data/lib/rbbt/sources/MSigDB.rb

@@ -0,0 +1,57 @@
+require 'rbbt-util'
+require 'rbbt/resource'
+require 'rbbt/sources/organism'
+
+module MSigDB
+  extend Resource
+  self.subdir = 'share/databases/MSigDB'
+
+  def self.organism(org="Hsa")
+    Organism.default_code(org)
+  end
+
+  #self.search_paths = {}
+  #self.search_paths[:default] = :lib
+
+  MSigDB.claim MSigDB['.source/all_sets.zip'], :proc do |filename|
+    raise "Download the 'ZIPed file set' from http://software.broadinstitute.org/gsea/downloads.jsp into #{filename}"
+  end
+
+  MSigDB.claim MSigDB.all_sets, :proc do |dirname|
+    zip_file = MSigDB['.source/all_sets.zip'].produce.find
+    TmpFile.with_dir do |tmpdir|
+      Misc.unzip_in_dir(zip_file, tmpdir)
+      Path.setup(tmpdir)
+      tmpdir.glob('**/*symbols.gmt').each do |file|
+        name_parts = File.basename(file).split(".")
+        base_name = name_parts[0..-5] * "_"
+        dumper = TSV::Dumper.new :key_field => "MSigDB Geneset ID", :fields => ["Associated Gene Name"], :namespace => MSigDB.organism, :type => :flat
+        dumper.init
+        io = TSV.traverse file, :type => :array, :into => dumper do |line|
+          name, url, *genes = line.split("\t")
+          [name, genes]
+        end
+        Open.write(dirname[base_name], io.stream)
+      end
+    end
+  end
+end
+
+if defined? Entity
+
+  module MSigDBGeneSet
+    extend Entity
+    self.format= "MSigDB Geneset ID"
+
+    property :name => :single2array do
+      self.downcase.gsub("_",' ')
+    end
+
+    property :genes => :single2array do
+      @@pathway_genes ||= MSigDB.all_sets.msigdb.tsv :persist => true
+      genes = @@pathway_genes[self]
+      Gene.setup(genes, "Associated Gene Name", MSigDB.organism).ensembl
+    end
+  end
+end
+

data/lib/rbbt/sources/uniprot.rb

@@ -72,7 +72,7 @@ module UniProt
   def self.get_uniprot_sequence(uniprotids)
     _array = Array === uniprotids
 
-    uniprotids = [uniprotids] unless Array === uniprotids
+    uniprotids = [uniprotids] unless _array
     uniprotids = uniprotids.compact.collect{|id| id}
 
     result_files = FileCache.cache_online_elements(uniprotids, 'uniprot-sequence-{ID}') do |ids|

data/share/install/Organism/organism_helpers.rb

@@ -48,6 +48,10 @@ $biomart_transcript_biotype = [
   ["Ensembl Transcript Biotype", 'transcript_biotype'],
 ]
 
+$biomart_transcript_name = [
+  ["Ensembl Transcript Name", 'external_transcript_id'],
+]
+
 
 $biomart_protein_sequence = [
   ['Protein Sequence','peptide'],
@@ -442,6 +446,12 @@ file 'transcript_biotype' do |t|
   Misc.sensiblewrite(t.name, biotype.to_s)
 end
 
+file 'transcript_name' do |t|
+  biotype = BioMart.tsv($biomart_db, $biomart_ensembl_transcript, $biomart_transcript_name, [], nil, :type => :single, :namespace => Thread.current['namespace'])
+
+  Misc.sensiblewrite(t.name, biotype.to_s)
+end
+
 file 'gene_pfam' do |t|
   pfam = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_pfam, [], nil, :type => :double, :namespace => Thread.current['namespace'])
 
@@ -700,3 +710,46 @@ file 'protein_sequence' => ["transcripts", "transcript_5utr", "transcript_3utr",
 
   Misc.sensiblewrite(t.name, protein_sequence.to_s)
 end
+
+file 'ensembl2uniprot' => ["protein_sequence", "protein_identifiers"] do |t|
+  ensp2unis  = TSV.open(File.expand_path('./protein_identifiers'), :key_field => "Ensembl Protein ID", :fields => ["UniProt/SwissProt Accession"], :type => :flat, :merge => true, :unnamed => true)
+  dumper = TSV::Dumper.new :key_field => "Ensembl Protein ID", :fields => ["UniProt/SwissProt Accession"], :namespace => Thread.current['namespace'], :type => :single
+  dumper.init
+  require 'rbbt/sources/uniprot'
+  TSV.traverse File.expand_path('./protein_sequence'), :into => dumper, :cpus => 20, :bar => true do |ensp,ensp_seq|
+    ensp = ensp.first if Array === ensp
+    unis = ensp2unis[ensp]
+    next if unis.nil? or unis.empty?
+    uni_seqs = UniProt.get_uniprot_sequence(unis)
+    best_uni = unis.zip(uni_seqs).sort_by do |uni,uni_seq|
+      (ensp_seq.length - uni_seq.length).abs
+    end.first.first
+    [ensp, best_uni]
+  end
+  Misc.sensiblewrite(t.name, dumper.stream)
+end
+
+file 'uniprot2ensembl' => ["protein_sequence", "protein_identifiers"] do |t|
+  uni2ensps  = TSV.open(File.expand_path('./protein_identifiers'), :fields => ["Ensembl Protein ID"], :key_field => "UniProt/SwissProt Accession", :type => :flat, :merge => true, :unnamed => true)
+  ensp2seq  = TSV.open(File.expand_path('./protein_sequence'), :unnamed => true)
+  dumper = TSV::Dumper.new :fields => ["Ensembl Protein ID"], :key_field => "UniProt/SwissProt Accession", :namespace => Thread.current['namespace'], :type => :single
+  dumper.init
+  require 'rbbt/sources/uniprot'
+  all_uni  = TSV.open(File.expand_path('./protein_identifiers'), :key_field => "UniProt/SwissProt Accession", :fields => [], :type => :double, :merge => true, :unnamed => true).keys.compact.reject{|u| u.empty?}
+  TSV.traverse all_uni, :into => dumper, :cpus => 1, :bar => true do |uni|
+    uni = uni.first if Array === uni
+    uni_seq = UniProt.get_uniprot_sequence(uni)
+    ensps = uni2ensps[uni]
+    next if ensps.nil? or ensps.empty?
+    best_ensp = ensps.sort_by do |ensp|
+      ensp_seq = ensp2seq[ensp]
+      if ensp_seq
+        (ensp_seq.length - uni_seq.length).abs
+      else
+        uni_seq.length
+      end
+    end.first
+    [uni, best_ensp]
+  end
+  Misc.sensiblewrite(t.name, dumper.stream)
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-sources
 version: !ruby/object:Gem::Version
-  version: 3.0.36
+  version: 3.0.37
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-11-06 00:00:00.000000000 Z
+date: 2015-11-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rbbt-util
@@ -92,6 +92,7 @@ files:
 - lib/rbbt/sources/COSTART.rb
 - lib/rbbt/sources/CTCAE.rb
 - lib/rbbt/sources/HPRD.rb
+- lib/rbbt/sources/MSigDB.rb
 - lib/rbbt/sources/NCI.rb
 - lib/rbbt/sources/PSI_MI.rb
 - lib/rbbt/sources/STITCH.rb