rbbt-sources 3.0.33 → 3.0.34

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 1f7fdfe0006f7704b66f77aadeed785f2cd1ec4a
-  data.tar.gz: bda1ec6492f1dcb961d0bc5b1b2775a78a19c16c
+  metadata.gz: 70e1df68dd361e27ccf4fa97f3d1566843aa9782
+  data.tar.gz: 700991e8dbbd83e10c956960d530692da3feff7a
 SHA512:
-  metadata.gz: aab537323d616bf284b5f54231179a808bb054d50f60fce8afd3a7b596aee4925063f737bdbc1d87ba749e786bbfec7d45356995b35c38414a614c049a37810d
-  data.tar.gz: dbb2e870406ea8cb7bd5bc7571b64e9c57841cb364d1553165a8510612607079917e054794e448fcdb8b70a81066c6c469463d774195a522d9bce431e02c296c
+  metadata.gz: 2f3eae3226db83878bb5fd5b58f7562ab38f3a7f36fbc4dc94752e4df0793834d9d2f451dcd0557275332e7ae81c2ec6ef10fde1f139f4eae089b8887bcb998d
+  data.tar.gz: 0c3e5d7294eaf73ce9c192dfa3073921ab5961936f87eb1578dc38b94b43147d559475f019a3c9294314fa603bd754dcba82db4040cd1f46180f0b5edf648df6

data/lib/rbbt/sources/go.rb

@@ -57,23 +57,32 @@ module GO
     end
   end
 
-  def self.id2ancestors(id)
+  def self.id2ancestors_by_type(id, type='is_a')
     if id.kind_of? Array
       info.values_at(*id).
-        select{|i| ! i['is_a'].nil?}.
-        collect{|i| i['is_a'].collect{|id| 
-        id.match(/(GO:\d+)/)[1] if id.match(/(GO:\d+)/)
-      }.compact
-      }
+        select{|i| ! i[type].nil?}.
+        collect{|i| 
+          res = i[type]
+          res = [res] unless Array === res
+          res.collect{|id| 
+            id.match(/(GO:\d+)/)[1] if id.match(/(GO:\d+)/)
+          }.compact
+        }
     else
-      return [] if id.nil? or info[id].nil? or info[id]['is_a'].nil?
-      info[id]['is_a'].
+      return [] if id.nil? or info[id].nil? or info[id][type].nil?
+      res = info[id][type]
+      res = [res] unless Array === res
+      res.
         collect{|id| 
         id.match(/(GO:\d+)/)[1] if id.match(/(GO:\d+)/)
       }.compact
     end
   end
 
+  def self.id2ancestors(id)
+    id2ancestors_by_type(id, 'is_a') + id2ancestors_by_type(id, 'relationship')
+  end
+
   def self.id2namespace(id)
     self.init unless info
     if id.kind_of? Array
@@ -89,10 +98,10 @@ module GO
     return ancestors if FalseClass === valid
     valid_ancestors = ancestors & valid
     return valid_ancestors if valid_ancestors.any?
-    valid_ancestors.inject([]) do |acc,ancestor|
+    ancestors.inject([]) do |acc,ancestor|
       valid_a = ancestors_in ancestor, valid
       acc = acc + valid_a
-    end
+    end.uniq
   end
 
   def self.group_genes(list, valid = nil)
@@ -100,10 +109,12 @@ module GO
 
     compartment_leaves = {}
     list.zip(list.go_cc_terms).each do |gene,terms|
+      terms = [] if terms.nil?
+      valid_terms = terms & valid
       valid_terms = terms.collect{|term| 
         (valid.include?(term) ? term : ancestors_in(term, valid))
       }.flatten
-      valid_terms - GOTerm.setup(valid_terms).flat_ancestry.flatten
+      #valid_terms - GOTerm.setup(valid_terms).flat_ancestry.flatten
       valid_terms.each do |term|
         compartment_leaves[term] ||= []
         compartment_leaves[term].push(gene)
@@ -112,7 +123,6 @@ module GO
 
     groups = {}
     while compartment_leaves.length > 1
-
       # Group common
       group = false
       new_compartment_leaves = {}
@@ -131,6 +141,7 @@ module GO
       if group == false
         new_compartment_leaves = {}
         final = compartment_leaves.keys
+        compartment_leaves
         compartment_leaves.each do |c,l|
           final = final - GOTerm.setup(c.dup).flat_ancestry
         end
@@ -139,10 +150,9 @@ module GO
           if final.include? c
             valid_an = ancestors_in c, valid
             valid_an.each do |ancestor|
-              ancestor = valid.first
-              next if ancestor.nil?
               new_compartment_leaves[ancestor] ||= []
-              new_compartment_leaves[ancestor].concat(l)
+              new_compartment_leaves[ancestor].push(c)
+              groups[ancestor].push(c) if groups[ancestor]
             end
           else
             new_compartment_leaves[c] = l
@@ -153,7 +163,7 @@ module GO
     end
     ng = {}
     groups.keys.reverse.each do |k|
-      ng[k] = {items: groups[k], id: k, name: id2name(k)}
+      ng[k] = {items: groups[k].uniq, id: k, name: id2name(k)}
     end
     ng
   end

data/share/install/Organism/organism_helpers.rb

@@ -44,6 +44,10 @@ $biomart_transcript_5utr = [
   ["5' UTR", '5utr'],
 ]
 
+$biomart_transcript_biotype = [
+  ["Ensembl Transcript Biotype", 'transcript_biotype'],
+]
+
 
 $biomart_protein_sequence = [
   ['Protein Sequence','peptide'],
@@ -432,6 +436,12 @@ file 'gene_biotype' do |t|
   Misc.sensiblewrite(t.name, biotype.to_s)
 end
 
+file 'transcript_biotype' do |t|
+  biotype = BioMart.tsv($biomart_db, $biomart_ensembl_transcript, $biomart_transcript_biotype, [], nil, :type => :single, :namespace => Thread.current['namespace'])
+
+  Misc.sensiblewrite(t.name, biotype.to_s)
+end
+
 file 'gene_pfam' do |t|
   pfam = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_pfam, [], nil, :type => :double, :namespace => Thread.current['namespace'])
 

data/test/rbbt/sources/test_go.rb

@@ -9,65 +9,45 @@ require 'test/unit'
 
 
 class TestGo < Test::Unit::TestCase
-  def _test_go
+  def test_go
     assert_match('vacuole inheritance',GO::id2name('GO:0000011'))
     assert_equal(['vacuole inheritance','alpha-glucoside transport'], GO::id2name(['GO:0000011','GO:0000017']))
   end
 
-  def _test_ancestors
+  def test_ancestors
     assert GO.id2ancestors('GO:0000001').include? 'GO:0048308'
   end
 
-  def _test_namespace
+  def test_namespace
     assert_equal 'biological_process', GO.id2namespace('GO:0000001')
   end
 
-  def _test_ancestors
+  def test_ancestors
     term = GOTerm.setup("GO:0005634")
   end
 
-  def _test_ancestry
+  def test_ancestry
     term = GOTerm.setup("GO:0005634")
     term.ancestry.include? "GO:0005634"
   end
 
-  def _test_ancestry_in
-    term = GOTerm.setup("GO:0005634")
+  def test_ancestors_in
+    term = GOTerm.setup("GO:0005730")
     valid = %w(GO:0005886 GO:0005634 GO:0005730 GO:0005829)
+    iii term.name
     iii GO.ancestors_in(term, valid)
   end
 
   def test_groups
-    list = Gene.setup(%w(ENSG00000009413
-                        ENSG00000038295
-                        ENSG00000038427
-                        ENSG00000047457
-                        ENSG00000058668
-                        ENSG00000065361
-                        ENSG00000070778
-                        ENSG00000072364
-                        ENSG00000073711
-                        ENSG00000075420
-                        ENSG00000088387
-                        ENSG00000096384
-                        ENSG00000100345
-                        ENSG00000102804
-                        ENSG00000102910
-                        ENSG00000103657
-                        ENSG00000104043
-                        ENSG00000106772
-                        ENSG00000107186
-                        ENSG00000108262
-                        ENSG00000261163
-                        ENSG00000263077
-                        ENSG00000101654
-                        ENSG00000111012), :organism => Organism.default_code("Hsa"))
+    list = Gene.setup(%w(FBXW7 SP140 LHX2 KIF23),
+                      "Associated Gene Name", Organism.default_code("Hsa"))
 
     valid = %w(GO:0005886 GO:0005634 GO:0005730 GO:0005829 )
+    valid = %w(GO:0005634 GO:0005730)
     iii GO.group_genes(list, valid)
   end
 
-  def _test_nucleolus
+  def test_nucleolus
     nuo = "GO:0005730"
     nu = "GO:0005634"
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-sources
 version: !ruby/object:Gem::Version
-  version: 3.0.33
+  version: 3.0.34
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-09-30 00:00:00.000000000 Z
+date: 2015-10-09 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rbbt-util