rbbt-sources 3.0.32 → 3.0.33

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: f704985c78e876aa8b63b7cc2367ae788bc35f61
-  data.tar.gz: 8e13c2f15915d79206b611196734fbf254488966
+  metadata.gz: 1f7fdfe0006f7704b66f77aadeed785f2cd1ec4a
+  data.tar.gz: bda1ec6492f1dcb961d0bc5b1b2775a78a19c16c
 SHA512:
-  metadata.gz: dd4ce9e020394422ad53405c60bd56824135014aed64374c6b8c54321d4a9f8e728c6d0345d80d08b1009fe74be5959b3dd9bfa5d8b524385b8f394c1400fc54
-  data.tar.gz: b9db1418b814865ae52f6600e3a634af63fba02b4b3f652257312f6645d484cb4b21168a8d38e2c94cb99ad399767a6d590b2321065d44638e67796f4d40f753
+  metadata.gz: aab537323d616bf284b5f54231179a808bb054d50f60fce8afd3a7b596aee4925063f737bdbc1d87ba749e786bbfec7d45356995b35c38414a614c049a37810d
+  data.tar.gz: dbb2e870406ea8cb7bd5bc7571b64e9c57841cb364d1553165a8510612607079917e054794e448fcdb8b70a81066c6c469463d774195a522d9bce431e02c296c

data/lib/rbbt/sources/go.rb

@@ -1,4 +1,4 @@
-require 'rbbt'
+require 'rbbt-util'
 require 'rbbt/resource'
 require 'rbbt/persist/tsv'
 
@@ -83,6 +83,80 @@ module GO
       info[id]['namespace']
     end
   end
+
+  def self.ancestors_in(term, valid)
+    ancestors = id2ancestors(term)
+    return ancestors if FalseClass === valid
+    valid_ancestors = ancestors & valid
+    return valid_ancestors if valid_ancestors.any?
+    valid_ancestors.inject([]) do |acc,ancestor|
+      valid_a = ancestors_in ancestor, valid
+      acc = acc + valid_a
+    end
+  end
+
+  def self.group_genes(list, valid = nil)
+    valid = %w(GO:0005886 GO:0005634 GO:0005730 GO:0005829) if valid.nil?
+
+    compartment_leaves = {}
+    list.zip(list.go_cc_terms).each do |gene,terms|
+      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.each do |term|
+        compartment_leaves[term] ||= []
+        compartment_leaves[term].push(gene)
+      end
+    end
+
+    groups = {}
+    while compartment_leaves.length > 1
+
+      # Group common
+      group = false
+      new_compartment_leaves = {}
+      compartment_leaves.each do |c,l|
+        if l.length > 1
+          groups[c] = l
+          group = true
+          new_compartment_leaves[c] = [c]
+        else
+          new_compartment_leaves[c] = l
+        end
+      end
+      compartment_leaves = new_compartment_leaves
+
+      # Pull-up leaves
+      if group == false
+        new_compartment_leaves = {}
+        final = compartment_leaves.keys
+        compartment_leaves.each do |c,l|
+          final = final - GOTerm.setup(c.dup).flat_ancestry
+        end
+
+        compartment_leaves.each do |c,l|
+          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)
+            end
+          else
+            new_compartment_leaves[c] = l
+          end
+        end
+        compartment_leaves = new_compartment_leaves
+      end
+    end
+    ng = {}
+    groups.keys.reverse.each do |k|
+      ng[k] = {items: groups[k], id: k, name: id2name(k)}
+    end
+    ng
+  end
 end
 
 if defined? Entity 
@@ -112,6 +186,31 @@ if defined? Entity
       description
     end
 
+    property :ancestors => :single2array do
+      GOTerm.setup(GO.id2ancestors(self))
+    end
+
+    property :ancestor => :single2array do
+      ancestors.first
+    end
+
+    property :ancestry => :single2array do
+      ancestry = {}
+      self.ancestors.each do |ancestor|
+        prev = ancestor.ancestry
+        ancestry[ancestor] = prev
+      end
+      ancestry
+    end
+
+    property :flat_ancestry => :single2array do
+      ancestry = []
+      self.ancestors.each do |ancestor|
+        prev = ancestor.flat_ancestry
+        ancestry = ancestry + [ancestor] + prev
+      end
+      ancestry
+    end
   end
 
   if defined? Gene and Entity === Gene
@@ -131,7 +230,7 @@ if defined? Entity
       property :go_mf_terms => :array2single do 
         @go_mf_terms ||= Organism.gene_go_mf(organism).tsv(:persist => true, :key_field => "Ensembl Gene ID", :fields => ["GO ID"], :type => :flat, :merge => true, :namespace => organism).chunked_values_at self.ensembl
       end
- 
+
     end
   end
 end

data/test/rbbt/sources/test_go.rb

@@ -1,21 +1,76 @@
 require File.expand_path(File.dirname(__FILE__) + '/../../test_helper')
 
+require 'rbbt/workflow'
+Workflow.require_workflow "Genomics"
+
+require 'rbbt/entity/gene'
 require 'rbbt/sources/go'
 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
+    term = GOTerm.setup("GO:0005634")
+  end
+
+  def _test_ancestry
+    term = GOTerm.setup("GO:0005634")
+    term.ancestry.include? "GO:0005634"
+  end
+
+  def _test_ancestry_in
+    term = GOTerm.setup("GO:0005634")
+    valid = %w(GO:0005886 GO:0005634 GO:0005730 GO:0005829)
+    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"))
+
+    valid = %w(GO:0005886 GO:0005634 GO:0005730 GO:0005829 )
+    iii GO.group_genes(list, valid)
+  end
+
+  def _test_nucleolus
+    nuo = "GO:0005730"
+    nu = "GO:0005634"
+  end
 end
 
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-sources
 version: !ruby/object:Gem::Version
-  version: 3.0.32
+  version: 3.0.33
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-07-06 00:00:00.000000000 Z
+date: 2015-09-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rbbt-util