mesh-medical-subject-headings 1.2.2 → 1.3.0

data/test/mesh_test.rb

@@ -58,6 +58,20 @@ module MESH
       assert_includes mh.tree_numbers, 'C20.111.163'
     end
 
+    it 'should have the correct root letters' do
+      mh = MESH::Mesh.find('D000224')
+      assert_equal ['C'], mh.roots
+      mh = MESH::Mesh.find('D064946')
+      assert_equal ['H', 'N'], mh.roots
+    end
+
+    it 'should have the correct descriptor class' do
+      mh = MESH::Mesh.find('D000224')
+      assert_equal :topical_descriptor, mh.descriptor_class
+      mh = MESH::Mesh.find('D005260')
+      assert_equal :check_tag, mh.descriptor_class
+    end
+
     it 'should have the correct original heading' do
       mh = MESH::Mesh.find('D000224')
       assert_equal 'Addison Disease', mh.original_heading
@@ -141,6 +155,10 @@ module MESH
       assert_includes parent.children, child4
     end
 
+    it 'should have the correct siblings' do
+      skip
+    end
+
     it 'should match on conditions for original_heading' do
       mh = MESH::Mesh.find('D001471')
       assert mh.matches(original_heading: /^Barrett Esophagus$/)
@@ -161,6 +179,29 @@ module MESH
       refute mh.matches(entries: /Foo/)
     end
 
+    it 'should match on descriptor class' do
+      mh = MESH::Mesh.find('D000224')
+      assert mh.matches(descriptor_class: :topical_descriptor)
+      refute mh.matches(descriptor_class: :check_tag)
+      mh = MESH::Mesh.find('D005260')
+      assert mh.matches(descriptor_class: :check_tag)
+      refute mh.matches(descriptor_class: :topical_descriptor)
+    end
+
+    it 'should match on conditions for tree numbers' do
+      mh = MESH::Mesh.find('D001471')
+      assert mh.matches(tree_numbers: /C06\.198\.102/)
+      assert mh.matches(tree_numbers: /^C06\.198\.102$/)
+      assert mh.matches(tree_numbers: /^C06/)
+      assert mh.matches(tree_numbers: /\.198\./)
+      assert mh.matches(tree_numbers: /^C06\.405\.117\.102$/)
+    end
+
+    it 'should not match on incorrect conditions for tree numbers' do
+      mh = MESH::Mesh.find('D001471')
+      refute mh.matches(tree_numbers: /Foo/)
+    end
+
     it 'should match on conditions for summary' do
       mh = MESH::Mesh.find('D001471')
       assert mh.matches(summary: /the lower ESOPHAGUS resulting from chronic acid reflux \(ESOPHAGITIS, REFLUX\)\./)
@@ -197,15 +238,26 @@ module MESH
     end
 
     it 'should match headings that occur in given text' do
-      expected_ids = %w(D001491 D001769 D001792 D001842 D001853 D002470 D002477 D002648 D002875 D002965 D002999 D003062 D003561 D003593 D003643 D004194 D004314 D004813 D004912 D005091 D005123 D005293 D005333 D005385 D005544 D005796 D006128 D006225 D006309 D006321 D006331 D006664 D007107 D007223 D007231 D007239 D007246 D007938 D008099 D008168 D008214 D008423 D008533 D008607 D008722 D009035 D009055 D009132 D009154 D009190 D009196 D009369 D009666 D010372 D010641 D011153 D012008 D012106 D012146 D012306 D012307 D012380 D012680 D012867 D013534 D013577 D013601 D013812 D013921 D013961 D014034 D014157 D014171 D014314 D014960 D015032 D015994 D015995 D016424 D016433 D017584 D017668 D018387 D018388 D019021 D019070 D019368 D019369 D032882 D036801 D038042 D041905 D052016 D055016)
+      expected_ids = %w(D001491 D001769 D001792 D001853 D002470 D002477 D002648 D002965 D002999 D003561 D003593 D003643 D004194 D004314 D004813 D004912 D005091 D005123 D005293 D005333 D005385 D005544 D005796 D006128 D006225 D006309 D006321 D006331 D006405 D007107 D007223 D007231 D007239 D007246 D007938 D007947 D008099 D008168 D008214 D008423 D008533 D008607 D008722 D009035 D009055 D009132 D009154 D009190 D009196 D009369 D009666 D010372 D010641 D011153 D012008 D012106 D012146 D012306 D012307 D012380 D012680 D012867 D013534 D013601 D013812 D013921 D013961 D014034 D014157 D014171 D014960 D015032 D015470 D015994 D015995 D016424 D016433 D017584 D017668 D018387 D018388 D019021 D019070 D019368 D019369 D032882 D036801 D038042 D041905 D052016 D054198 D055016)
       expected = expected_ids.map { |id| MESH::Mesh.find(id) }
       matches = MESH::Mesh.match_in_text(@example_text)
       actual = matches.map { |match| match[:heading] }.uniq
-      assert_equal expected, actual
+      assert_equal expected.sort, actual.sort
+    end
+
+    it 'should sort based on unique id' do
+      skip
+    end
+
+    it 'should only match the most specific matches in given text' do
+      expected = MESH::Mesh.find('D054144')
+      actual = MESH::Mesh.match_in_text('Diastolic Heart Failure')
+      assert_equal 1, actual.length
+      assert_equal expected, actual.first[:heading]
     end
 
     it 'should only match useful headings that occur in given text' do
-      expected_ids = %w(D001491 D001769 D001792 D001842 D001853 D002470 D002648 D002875 D002965 D003062 D003561 D003593 D003643 D004194 D004314 D004813 D004912 D005091 D005123 D005293 D005333 D005385 D005544 D005796 D006128 D006225 D006309 D006321 D006331 D007107 D007231 D007239 D007938 D008099 D008168 D008214 D008423 D008607 D008722 D009035 D009055 D009132 D009154 D009190 D009196 D009369 D009666 D010372 D010641 D011153 D012008 D012106 D012146 D012306 D012307 D012380 D012680 D012867 D013534 D013577 D013601 D013812 D013921 D013961 D014034 D014157 D014171 D014314 D015032 D015994 D015995 D016424 D017584 D017668 D018387 D018388 D019021 D019070 D019368 D019369 D032882 D036801 D038042 D041905 D052016)
+      expected_ids = %w(D001491 D001769 D001792 D001853 D002470 D002648 D002875 D002965 D003561 D003593 D003643 D004194 D004314 D004813 D004912 D005091 D005123 D005293 D005333 D005385 D005544 D005796 D006128 D006225 D006309 D006321 D006331 D006405 D007107 D007231 D007239 D007938 D007947 D008099 D008168 D008214 D008423 D008607 D008722 D009035 D009055 D009132 D009154 D009190 D009196 D009369 D009666 D010372 D010641 D011153 D012008 D012106 D012146 D012306 D012307 D012380 D012680 D012867 D013534 D013601 D013812 D013921 D013961 D014034 D014157 D014171 D015032 D015470 D015994 D015995 D016424 D017584 D017668 D018387 D018388 D019021 D019070 D019368 D019369 D032882 D036801 D038042 D041905 D052016 D054198)
 
       not_useful_ids = %w(D007246 D002477 D014960 D008533 D016433 D006664 D055016 D002999 D007223)
       begin
@@ -234,6 +286,26 @@ module MESH
       assert_equal MESH::Mesh.find('D007938'), matches[0][:heading]
     end
 
+    it 'should return no matches when given nil text' do
+      assert_equal [], MESH::Mesh.match_in_text(nil)
+    end
+
+    it 'should only match uppercase entries with uppercase text' do
+      text = 'Lorem amet, consectetur adipiscing elit. Donec pretium ATP leo diam, quis adipiscing purus bibendum.'
+      matches = MESH::Mesh.match_in_text(text)
+      assert_equal 1, matches.length
+      assert_equal MESH::Mesh.find('D000255'), matches[0][:heading]
+      text = 'Lorem ipsum consectetur adipiscing elit. Donec pretium atp leo diam, quis adipiscing purus bibendum.'
+      assert_equal [], MESH::Mesh.match_in_text(text)
+    end
+
+    it 'should match anglicised terms in text' do
+      text = 'Lorem ipsum dolor sit amet, consectetur adipiscing elit. Donec pretium leo diam, quis adipiscing purus bibendum eu leukaemia'
+      matches = MESH::Mesh.match_in_text(text)
+      assert_equal 1, matches.length
+      assert_equal MESH::Mesh.find('D007938'), matches[0][:heading]
+    end
+
     it 'should allow headings to be marked as not useful' do
       mh = MESH::Mesh.find('D055550')
       mh.useful = true
@@ -244,9 +316,29 @@ module MESH
       assert mh.useful
     end
 
+    it 'should allow headings to be found with a where() match on original_heading' do
+      expected = [MESH::Mesh.find('D003561'), MESH::Mesh.find('D016238')]
+      actual = MESH::Mesh.where(original_heading: /^Cyta/)
+      assert_equal expected, actual
+    end
+
+    it 'should match on entries in where()' do
+      expected_ids = %w( D002397 D003064 D003400 D003532 D004284 D004289 D004555 D005412 D006054 D006196 D007059 D007497 D007695 D009990 D010473 D012091 D012487 D012758 D013215 D015027 D020410 D023721 D023761 D023781 D024541 D029961 D037401 D037462 D048251 D049832 D052656 D057096 )
+      expected = expected_ids.map { |id| MESH::Mesh.find(id) }
+      actual = MESH::Mesh.where(entries: /fish/)
+      assert_equal expected, actual
+    end
+
+    it 'should match on tree numbers in where()' do
+      expected_ids = %w( D000005 D001415 D010388 D013909 )
+      expected = expected_ids.map { |id| MESH::Mesh.find(id) }
+      actual = MESH::Mesh.where(tree_numbers: /^A01\.923\.[0-9]{3}$/)
+      assert_equal expected, actual
+    end
+
     it 'should match on useful in where' do
+      expected = [MESH::Mesh.find('D012000'), MESH::Mesh.find('D064906'), MESH::Mesh.find('D064966')]
       begin
-        expected = [MESH::Mesh.find('D012000'), MESH::Mesh.find('D064906'), MESH::Mesh.find('D064966')]
         expected.each { |mh| mh.useful = false }
         actual = MESH::Mesh.where(useful: false)
         assert_equal expected, actual
@@ -255,6 +347,17 @@ module MESH
       end
     end
 
+    it 'should match on descriptor class in where' do
+      expected = [MESH::Mesh.find('D012000'), MESH::Mesh.find('D064906'), MESH::Mesh.find('D064966')]
+      begin
+        expected.each { |mh| mh.descriptor_class = :foo }
+        actual = MESH::Mesh.where(descriptor_class: :foo)
+        assert_equal expected, actual
+      ensure
+        expected.each { |mh| mh.descriptor_class = :topical_descriptor }
+      end
+    end
+
     it 'should only include useful headings in .each' do
       begin
         MESH::Mesh.each do |mh|
@@ -276,19 +379,75 @@ module MESH
       end
     end
 
+    it 'should know its deepest position in the tree' do
+      #single tree numbers
+      assert_equal 1, MESH::Mesh.find('D002319').deepest_position
+      assert_equal 2, MESH::Mesh.find('D001808').deepest_position
+      assert_equal 3, MESH::Mesh.find('D001158').deepest_position
+      assert_equal 4, MESH::Mesh.find('D001981').deepest_position
+    end
+
+    it 'should know its shallowest position in the tree' do
+      #single tree numbers
+      assert_equal 1, MESH::Mesh.find('D002319').shallowest_position
+      assert_equal 2, MESH::Mesh.find('D001808').shallowest_position
+      assert_equal 3, MESH::Mesh.find('D001158').shallowest_position
+      assert_equal 4, MESH::Mesh.find('D001981').shallowest_position
+    end
+
+    it 'should know if one heading is the descendant of another' do
+      parent = MESH::Mesh.find('D002319')
+      child = MESH::Mesh.find('D001808')
+      grandchild = MESH::Mesh.find('D001158')
+      great_grandchild = MESH::Mesh.find('D001981')
+      unrelated = MESH::Mesh.find('D008091')
+
+      refute parent.has_descendant(parent), 'should not consider itself a desecendant'
+      assert parent.has_descendant(child), "should consider child #{child.inspect} a descendant of #{parent.inspect} in #{parent.children}"
+      assert parent.has_descendant(grandchild), "should consider grandchild #{grandchild.inspect} a descendant #{parent.inspect}"
+      assert parent.has_descendant(great_grandchild), "should consider great grandchild #{great_grandchild.inspect} a descendant #{parent.inspect}"
+      refute parent.has_descendant(unrelated), 'should not consider an unrelated heading a descendant'
+    end
+
+    it 'should know if one heading is the ancestor of another' do
+      child = MESH::Mesh.find('D001981')
+      parent = MESH::Mesh.find('D001158')
+      grandparent = MESH::Mesh.find('D001808')
+      great_grandparent = MESH::Mesh.find('D002319')
+      unrelated = MESH::Mesh.find('D008091')
+
+      refute child.has_ancestor(child), 'should not consider itself an ancestor'
+      assert child.has_ancestor(parent), "should consider parent #{parent.inspect} an ancestor of #{child.inspect} in #{child.parents}"
+      assert child.has_ancestor(grandparent), "should consider grandparent #{grandparent.inspect} an ancestor #{child.inspect}"
+      assert child.has_ancestor(great_grandparent), "should consider great grandparent #{great_grandparent.inspect} an ancestor #{child.inspect}"
+      refute child.has_ancestor(unrelated), 'should not consider an unrelated heading an ancestor'
+    end
+
+    it 'should know if headings are siblings at the same level below a common parent' do
+      parent = MESH::Mesh.find_by_tree_number('C19.053.500')
+      child1 = MESH::Mesh.find_by_tree_number('C19.053.500.263')
+      child2 = MESH::Mesh.find_by_tree_number('C19.053.500.270')
+      child3 = MESH::Mesh.find_by_tree_number('C19.053.500.480')
+      child4 = MESH::Mesh.find_by_tree_number('C19.053.500.740')
+      unrelated = MESH::Mesh.find('D008091')
+      children = [child1, child2, child3, child4]
+
+      children.each { |c| refute parent.sibling?(c) }
+      children.each { |c| refute c.sibling?(parent) }
+
+      children.each { |c| assert child1.sibling?(c) unless c == child1 }
+      children.each { |c| assert c.sibling?(child1) unless c == child1 }
+
+      children.each { |c| refute unrelated.sibling?(c) }
+      children.each { |c| refute c.sibling?(unrelated) }
+    end
+
     it 'should override inspect to prevent issues in test diagnostics' do
       mh = MESH::Mesh.find('D001471')
-      expected = "#{mh.unique_id}, #{mh.original_heading}"
+      expected = "#{mh.unique_id}, #{mh.original_heading}, [#{mh.tree_numbers.join(',')}]"
       assert_equal expected, mh.inspect
     end
 
-    it 'should allow headings to be found with a where() match on original_heading' do
-      expected = [MESH::Mesh.find('D003561'), MESH::Mesh.find('D016238')]
-      actual = MESH::Mesh.where(original_heading: /^Cyta/)
-      assert_equal expected, actual
-    end
-    #MESH::Mesh.where(:entries, /.*Fish.*/)
-
     before do
 
       @example_text = 'Leukaemia in Downs Syndrome

data/test/test_helper.rb

@@ -9,8 +9,17 @@ $LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
 
 require "MESH"
 
+puts 'Configuring MESH::Mesh — this may take up to 10 seconds.'
 start = Time.now
 MESH::Mesh.configure(filename: File.expand_path('../../data/mesh_data_2014/d2014.bin.gz', __FILE__))
 finish = Time.now
 configuration_time = finish - start
-raise 'MESH::Mesh should configure in less than 10 seconds.' unless configuration_time < 10
+#raise 'MESH::Mesh should configure in less than 10 seconds.' unless configuration_time < 10
+
+puts 'Translating MESH::Mesh into English ;) — this may take up to 60 seconds.'
+start = Time.now
+MESH::Mesh.translate('en-GB', MESH::Translator.new(MESH::Translator.enus_to_engb))
+finish = Time.now
+configuration_time = finish - start
+puts "took #{configuration_time}"
+#raise 'MESH::Mesh should translate in less than 30 seconds.' unless configuration_time < 60

data/test/translator_test.rb

@@ -5,14 +5,18 @@ module MESH
   describe 'Testing MESH:Translator core functions' do
 
     it 'should translate a single word' do
-      tr = MESH::Translator.new
+      tr = MESH::Translator.new(MESH::Translator.enus_to_engb)
       assert_equal 'oesophagus', tr.translate('esophagus')
       assert_equal 'aluminium', tr.translate('aluminum')
       assert_equal 'gynaecology', tr.translate('gynecology')
+      tr = MESH::Translator.new(MESH::Translator.engb_to_enus)
+      assert_equal 'esophagus', tr.translate('oesophagus')
+      assert_equal 'aluminum', tr.translate('aluminium')
+      assert_equal 'gynecology', tr.translate('gynaecology')
     end
 
     it 'should translate within a body of text' do
-      tr = MESH::Translator.new
+      tr = MESH::Translator.new(MESH::Translator.enus_to_engb)
       input = 'a condition with damage to the lining of the lower esophagus resulting from chronic acid reflux (esophagitis, reflux). through the process of metaplasia, the squamous cells are replaced by a columnar epithelium with cells resembling those of the intestine or the salmon-pink mucosa of the stomach. barrett\'s columnar epithelium is a marker for severe reflux and precursor to adenocarcinoma of the esophagus.'
       expected = 'a condition with damage to the lining of the lower oesophagus resulting from chronic acid reflux (oesophagitis, reflux). through the process of metaplasia, the squamous cells are replaced by a columnar epithelium with cells resembling those of the intestine or the salmon-pink mucosa of the stomach. barrett\'s columnar epithelium is a marker for severe reflux and precursor to adenocarcinoma of the oesophagus.'
 
@@ -20,28 +24,28 @@ module MESH
     end
 
     it 'should match uppercase' do
-      tr = MESH::Translator.new
+      tr = MESH::Translator.new(MESH::Translator.enus_to_engb)
       input = 'A condition with damage to the lining of the lower ESOPHAGUS resulting from chronic acid reflux (ESOPHAGITIS, REFLUX). Through the process of metaplasia, the squamous cells are replaced by a columnar epithelium with cells resembling those of the INTESTINE or the salmon-pink mucosa of the STOMACH. Barrett\'s columnar epithelium is a marker for severe reflux and precursor to ADENOCARCINOMA of the esophagus.'
       expected = 'A condition with damage to the lining of the lower OESOPHAGUS resulting from chronic acid reflux (OESOPHAGITIS, REFLUX). Through the process of metaplasia, the squamous cells are replaced by a columnar epithelium with cells resembling those of the INTESTINE or the salmon-pink mucosa of the STOMACH. Barrett\'s columnar epithelium is a marker for severe reflux and precursor to ADENOCARCINOMA of the oesophagus.'
       assert_equal expected, tr.translate(input)
     end
 
     it 'should match title case' do
-      tr = MESH::Translator.new
+      tr = MESH::Translator.new(MESH::Translator.enus_to_engb)
       input = 'A condition with damage to the lining of the lower Esophagus resulting from chronic acid reflux (Esophagitis, REFLUX). Through the process of metaplasia, the squamous cells are replaced by a columnar epithelium with cells resembling those of the INTESTINE or the salmon-pink mucosa of the STOMACH. Barrett\'s columnar epithelium is a marker for severe reflux and precursor to ADENOCARCINOMA of the esophagus.'
       expected = 'A condition with damage to the lining of the lower Oesophagus resulting from chronic acid reflux (Oesophagitis, REFLUX). Through the process of metaplasia, the squamous cells are replaced by a columnar epithelium with cells resembling those of the INTESTINE or the salmon-pink mucosa of the STOMACH. Barrett\'s columnar epithelium is a marker for severe reflux and precursor to ADENOCARCINOMA of the oesophagus.'
       assert_equal expected, tr.translate(input)
     end
 
     it 'should not change the input string' do
-      tr = MESH::Translator.new
+      tr = MESH::Translator.new(MESH::Translator.enus_to_engb)
       input = 'esophagus'
       assert_equal 'oesophagus', tr.translate(input)
       assert_equal 'esophagus', input
     end
 
     it 'should maintain punctuation' do
-      tr = MESH::Translator.new
+      tr = MESH::Translator.new(MESH::Translator.enus_to_engb)
       input = 'Esophagus, Barrett'
       assert_equal 'Oesophagus, Barrett', tr.translate(input)
     end

data/tr_speed.rb

@@ -0,0 +1,8 @@
+require_relative 'lib/MESH'
+
+puts DateTime.now
+MESH::Mesh.configure(filename: 'data/mesh_data_2014/d2014.bin.gz')
+puts DateTime.now
+MESH::Mesh.translate('en-GB', MESH::Translator.new(MESH::Translator.enus_to_engb))
+puts DateTime.now
+

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: mesh-medical-subject-headings
 version: !ruby/object:Gem::Version
-  version: 1.2.2
+  version: 1.3.0
 platform: ruby
 authors:
-- mmmmmrob
+- Rob Styles
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-03-05 00:00:00.000000000 Z
+date: 2014-04-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -117,13 +117,16 @@ files:
 - data/mesh_data_2014/q2014.bin.gz
 - data/mesh_data_2014/useful_2014.tsv
 - lib/MESH.rb
+- lib/MESH/classifier.rb
 - lib/MESH/mesh.rb
 - lib/MESH/translator.rb
 - lib/MESH/version.rb
 - match.rb
+- test/classifier_test.rb
 - test/mesh_test.rb
 - test/test_helper.rb
 - test/translator_test.rb
+- tr_speed.rb
 homepage: ''
 licenses:
 - AGPL3
@@ -150,6 +153,7 @@ specification_version: 4
 summary: A ruby gem containing MeSH subject headings (https://www.nlm.nih.gov/mesh/)
   for use in classifying and entity recognition.
 test_files:
+- test/classifier_test.rb
 - test/mesh_test.rb
 - test/test_helper.rb
 - test/translator_test.rb