lazar dumpfiles implemented

3 files changed, 117 insertions, 119 deletions

diff --git a/fminer.rb b/fminer.rb
index 87ea4a4..22fc945 100644
--- a/fminer.rb
+++ b/fminer.rb
@@ -16,16 +16,16 @@ class FminerTest < Test::Unit::TestCase
     @dataset.delete(@@subjectid)
   end
 
-  def dump(method)
+  def dump
     @dataset = OpenTox::Dataset.find @dataset_uri, @@subjectid
-    @dumpfile = File.join(@dump_dir,method.to_s)+".yaml"
+    @dumpfile = File.join(@dump_dir,caller[0][/`.*'/][1..-2])+".yaml"
     File.open(@dumpfile,"w+"){|f| f.puts @dataset.to_yaml}
   end
 
   def test_bbrc
     feature = @@classification_training_dataset.features.keys.first
     @dataset_uri = OpenTox::Algorithm::Fminer::BBRC.new.run({:dataset_uri => @@classification_training_dataset.uri, :prediction_feature => feature, :subjectid => @@subjectid}).to_s
-    dump __method__
+    dump 
     assert_equal 52, @dataset.features.size
     cleanup
   end
@@ -33,7 +33,7 @@ class FminerTest < Test::Unit::TestCase
   def test_regression_bbrc
     feature = File.join @@regression_training_dataset.uri,"feature/LC50_mmol" 
     @dataset_uri = OpenTox::Algorithm::Fminer::BBRC.new.run({:dataset_uri => @@regression_training_dataset.uri, :prediction_feature => feature, :subjectid => @@subjectid, :feature_type=>"paths"}).to_s
-    dump __method__
+    dump
     assert_equal 219, @dataset.features.size
     cleanup
   end
@@ -41,7 +41,7 @@ class FminerTest < Test::Unit::TestCase
   def test_last
     feature = @@classification_training_dataset.features.keys.first
     @dataset_uri = OpenTox::Algorithm::Fminer::LAST.new.run({:dataset_uri => @@classification_training_dataset.uri, :prediction_feature => feature, :subjectid => @@subjectid}).to_s
-    dump __method__
+    dump
     assert_equal 23, @dataset.features.size
     cleanup
   end
@@ -54,7 +54,7 @@ class FminerTest < Test::Unit::TestCase
       "backbone" => true,
       "min_frequency" => 2,
        :subjectid => @@subjectid })
-    dump __method__
+    dump
     assert_equal 52, @dataset.features.size
     cleanup
   end
@@ -62,11 +62,10 @@ class FminerTest < Test::Unit::TestCase
 # Deactivated by AM because of efficiency problems (does not return)
 #  def test_regression_last
 #    feature = File.join @@regression_training_dataset.uri,"feature/LC50_mmol" 
-#    dataset_uri = OpenTox::Algorithm::Fminer::LAST.new.run({:dataset_uri => @@regression_training_dataset.uri, :prediction_feature => feature, :subjectid => @@subjectid}).to_s
-#    d =OpenTox::Dataset.new dataset_uri, @@subjectid
-#    d.load_features(@@subjectid)
+#    @dataset_uri = OpenTox::Algorithm::Fminer::LAST.new.run({:dataset_uri => @@regression_training_dataset.uri, :prediction_feature => feature, :subjectid => @@subjectid}).to_s
+#    dump
 #    assert_equal 4, d.features.size
-#    d.delete(@@subjectid)
+#    cleanup
 #  end
 
 end
diff --git a/lazar.rb b/lazar.rb
index b60794c..ea41f98 100644
--- a/lazar.rb
+++ b/lazar.rb
@@ -13,143 +13,138 @@ class LazarTest < Test::Unit::TestCase
 
   def setup
     @predictions = []
-    @models = []
+    @compounds = []
+    @files = []
+    @dump_dir = FileUtils.mkdir_p File.join(File.dirname(__FILE__),"dump",File.basename(__FILE__,".rb"))
+    FileUtils.mkdir_p File.join(File.dirname(__FILE__),"reference",File.basename(__FILE__,".rb"))
   end
 
-  def teardown
-    @predictions.each {|p| p.delete(@@subjectid)}
-    @models.each {|m| m.delete(@@subjectid)}
+  def dump(object,file)
+    @files << file
+    FileUtils.mkdir_p File.dirname(file)
+    File.open(file,"w+"){|f| f.puts object.to_yaml}
   end
 
-=begin
-=end
-  def test_create_regression_model
-    model_uri = OpenTox::Algorithm::Lazar.new.run({:dataset_uri => @@regression_training_dataset.uri, :subjectid => @@subjectid}).to_s
-    #puts model_uri
-    validate_owl model_uri,@@subjectid
-    lazar = OpenTox::Model::Lazar.find model_uri, @@subjectid
-    @models << lazar
-    compound = OpenTox::Compound.from_smiles("c1ccccc1NN")
-    prediction_uri = lazar.run(:compound_uri => compound.uri, :subjectid => @@subjectid).to_s
+  def create_model(params)
+    params[:subjectid] = @@subjectid
+    model_uri = OpenTox::Algorithm::Lazar.new.run(params).to_s
+    @model = OpenTox::Model::Lazar.find model_uri, @@subjectid
+    dump @model, File.join(@dump_dir,caller[0][/`.*'/][1..-2],"model")+".yaml"
+  end
+
+  def predict_compound(compound)
+    @compounds << compound
+    prediction_uri = @model.run(:compound_uri => compound.uri, :subjectid => @@subjectid)
     prediction = OpenTox::LazarPrediction.find(prediction_uri, @@subjectid)
     @predictions << prediction
-    assert_equal prediction.value(compound).round_to(3),0.378.round_to(3)
-    assert_equal prediction.confidence(compound).round_to(3), 0.276.round_to(3)
-    #assert_equal prediction.value(compound).round_to(4), 0.2847.round_to(4)
-    #assert_equal prediction.confidence(compound).round_to(4), 0.3223.round_to(4)
-    assert_equal prediction.neighbors(compound).size, 61
+    dump prediction, File.join(@dump_dir,caller[0][/`.*'/][1..-2],"compound_prediction")+@compounds.size.to_s+".yaml"
   end
 
-  def test_create_regression_prop_model
-    model_uri = OpenTox::Algorithm::Lazar.new.run({:dataset_uri => @@regression_training_dataset.uri, :subjectid => @@subjectid, :local_svm_kernel => "propositionalized"}).to_s
-    #puts model_uri
-    validate_owl model_uri,@@subjectid
-    lazar = OpenTox::Model::Lazar.find model_uri, @@subjectid
-    @models << lazar
-    assert_equal 219, lazar.features.size
-    compound = OpenTox::Compound.from_smiles("c1ccccc1NN")
-    prediction_uri = lazar.run(:compound_uri => compound.uri, :subjectid => @@subjectid).to_s
+  def predict_dataset(dataset)
+    prediction_uri = @model.run(:dataset_uri => dataset.uri, :subjectid => @@subjectid)
     prediction = OpenTox::LazarPrediction.find(prediction_uri, @@subjectid)
     @predictions << prediction
-    assert_equal prediction.value(compound).round_to(1),0.1.round_to(1)
-    assert_equal prediction.confidence(compound).round_to(3), 0.276.round_to(3)
-    #assert_equal prediction.value(compound).round_to(4), 0.2847.round_to(4)
-    #assert_equal prediction.confidence(compound).round_to(4), 0.3223.round_to(4)
-    assert_equal prediction.neighbors(compound).size, 61
+    dump prediction, File.join(@dump_dir,caller[0][/`.*'/][1..-2],"dataset_prediction")+".yaml"
   end
 
-  def test_classification_model
+  def cleanup # executed only when assertions succeed (teardown is called even when assertions fail)
+    validate_owl @model.uri
+    @files.each do |f|
+      reference = f.sub(/dump/,"reference")
+      FileUtils.mkdir_p File.dirname(reference)
+      FileUtils.cp f, reference
+      FileUtils.rm f
+    end
+    @predictions.each do |dataset|
+      validate_owl @model.uri
+      dataset.delete(@@subjectid)
+    end
+    @model.delete(@@subjectid)
+  end
 
-    # create model
-    model_uri = OpenTox::Algorithm::Lazar.new.run({:dataset_uri => @@classification_training_dataset.uri, :subjectid => @@subjectid}).to_s
-    validate_owl model_uri,@@subjectid
-    lazar = OpenTox::Model::Lazar.find model_uri, @@subjectid
-    @models << lazar
-    assert_equal lazar.features.size, 52
+=begin
+=end
+  def test_create_regression_model
+    create_model :dataset_uri => @@regression_training_dataset.uri
+    predict_compound OpenTox::Compound.from_smiles("c1ccccc1NN")
+    assert_equal 0.4.round_to(3), @predictions.first.value(@compounds.first).round_to(3)
+    assert_equal 0.276.round_to(3), @predictions.first.confidence(@compounds.first).round_to(3)
+    assert_equal 61, @predictions.first.neighbors(@compounds.first).size
+    cleanup
+  end
 
-    # single prediction
-    compound = OpenTox::Compound.from_smiles("c1ccccc1NN")
-    prediction_uri = lazar.run(:compound_uri => compound.uri, :subjectid => @@subjectid)
-    prediction = OpenTox::LazarPrediction.find(prediction_uri, @@subjectid)
-    @predictions << prediction
-    #puts prediction_uri
-    assert_equal prediction.value(compound), "false"
-    assert_equal prediction.confidence(compound).round_to(4), 0.3067.round_to(4)
-    assert_equal prediction.neighbors(compound).size, 14
+  def test_create_regression_prop_model
+    create_model :dataset_uri => @@regression_training_dataset.uri, :local_svm_kernel => "propositionalized"
+    predict_compound  OpenTox::Compound.from_smiles("c1ccccc1NN")
+    assert_equal 0.4.round_to(1), @predictions.first.value(@compounds.first).round_to(1)
+    assert_equal 0.276.round_to(3), @predictions.first.confidence(@compounds.first).round_to(3)
+    assert_equal 61, @predictions.first.neighbors(@compounds.first).size
+    assert_equal 219, @model.features.size
+    cleanup
+  end
 
+  def test_classification_model
+    create_model :dataset_uri => @@classification_training_dataset.uri
+    # single prediction
+    predict_compound OpenTox::Compound.from_smiles("c1ccccc1NN")
     # dataset activity
-    compound = OpenTox::Compound.from_smiles("CNN")
-    prediction_uri  = lazar.run(:compound_uri => compound.uri, :subjectid => @@subjectid)
-    prediction = OpenTox::LazarPrediction.find prediction_uri, @@subjectid
-    @predictions << prediction
-    assert !prediction.measured_activities(compound).empty?
-    assert_equal prediction.measured_activities(compound).first.to_s, "true"
-    assert prediction.value(compound).nil?
-
+    predict_compound OpenTox::Compound.from_smiles("CNN")
     # dataset prediction
-    test_dataset = OpenTox::Dataset.create_from_csv_file("data/multicolumn.csv", @@subjectid)
-    prediction = OpenTox::LazarPrediction.find lazar.run(:dataset_uri => test_dataset.uri, :subjectid => @@subjectid), @@subjectid
-    @predictions << prediction
-    assert_equal prediction.compounds.size, 4
-    compound = OpenTox::Compound.from_smiles "CC(=Nc1ccc2c(c1)Cc1ccccc21)O"
-    assert_equal prediction.value(compound), nil
-    assert_equal prediction.measured_activities(compound).first.to_s, "true"
+    predict_dataset OpenTox::Dataset.create_from_csv_file("data/multicolumn.csv", @@subjectid)
+    # assertions
+    # single prediction
+    assert_equal false, @predictions[0].value(@compounds[0])
+    assert_equal 0.3067.round_to(4), @predictions[0].confidence(@compounds[0]).round_to(4)
+    assert_equal 14, @predictions[0].neighbors(@compounds[0]).size
+    # dataset activity
+    assert !@predictions[1].measured_activities(@compounds[1]).empty?
+    assert_equal "true", @predictions[1].measured_activities(@compounds[1]).first.to_s
+    assert @predictions[1].value(@compounds[1]).nil?
+    # dataset prediction
+    c = OpenTox::Compound.from_smiles("CC(=Nc1ccc2c(c1)Cc1ccccc21)O")
+    assert_equal nil, @predictions[2].value(c)
+    assert_equal "true", @predictions[2].measured_activities(c).first.to_s
+    c = OpenTox::Compound.new("http://ot-dev.in-silico.ch/compound/InChI=1S/C2H4N4/c3-2-4-1-5-6-2/h1H,(H3,3,4,5,6)")
+    assert_equal false, @predictions[2].value(c)
+    # model
+    assert_equal 52, @model.features.size
+    cleanup
   end
 
   def test_classification_svm_model
 
-    # create model
-    model_uri = OpenTox::Algorithm::Lazar.new.run({:dataset_uri => @@classification_training_dataset.uri, :subjectid => @@subjectid, :prediction_algorithm => "local_svm_classification"}).to_s
-    lazar = OpenTox::Model::Lazar.find model_uri, @@subjectid
-    @models << lazar
-    assert_equal lazar.features.size, 52
+    create_model :dataset_uri => @@classification_training_dataset.uri, :prediction_algorithm => "local_svm_classification"
+    predict_compound OpenTox::Compound.from_smiles("c1ccccc1NN")
+    predict_dataset OpenTox::Dataset.create_from_csv_file("data/multicolumn.csv", @@subjectid)
 
-    # single prediction
-    compound = OpenTox::Compound.from_smiles("c1ccccc1NN")
-    prediction_uri = lazar.run(:compound_uri => compound.uri, :subjectid => @@subjectid)
-    prediction = OpenTox::LazarPrediction.find(prediction_uri, @@subjectid)
-    @predictions << prediction
-    assert_equal prediction.value(compound), "false"
-    assert_equal prediction.confidence(compound).round_to(4), 0.4131.round_to(4)
-    assert_equal prediction.neighbors(compound).size, 14
+    assert_equal false, @predictions[0].value(@compounds[0])
+    assert_equal 0.3067.round_to(4), @predictions[0].confidence(@compounds[0]).round_to(4)
+    assert_equal 14, @predictions[0].neighbors(@compounds[0]).size
 
-    # dataset prediction
-    test_dataset = OpenTox::Dataset.create_from_csv_file("data/multicolumn.csv", @@subjectid)
-    prediction = OpenTox::LazarPrediction.find lazar.run(:dataset_uri => test_dataset.uri, :subjectid => @@subjectid), @@subjectid
-    @predictions << prediction
-    assert_equal prediction.compounds.size, 4
-    compound = OpenTox::Compound.from_smiles "CC(=Nc1ccc2c(c1)Cc1ccccc21)O"
-    assert_equal prediction.value(compound), nil
-    assert_equal prediction.measured_activities(compound).first, true
+    c = OpenTox::Compound.new("http://ot-dev.in-silico.ch/compound/InChI=1S/C2H4N4/c3-2-4-1-5-6-2/h1H,(H3,3,4,5,6)")
+    assert_equal 4, @predictions[1].compounds.size
+    assert_equal false, @predictions[1].value(c)
+
+    assert_equal 52, @model.features.size
+    cleanup
  end
 
   def test_classification_svm_prop_model
 
+    create_model :dataset_uri => @@classification_training_dataset.uri, :prediction_algorithm => "local_svm_classification", :local_svm_kernel => "propositionalized"
+    predict_compound OpenTox::Compound.from_smiles("c1ccccc1NN")
+    predict_dataset OpenTox::Dataset.create_from_csv_file("data/multicolumn.csv", @@subjectid)
+    
+    assert_equal false, @predictions[0].value(@compounds[0])
+    assert_equal 0.3067.round_to(4), @predictions[0].confidence(@compounds[0]).round_to(4)
+    assert_equal 14, @predictions[0].neighbors(@compounds[0]).size
 
-    # create model
-    model_uri = OpenTox::Algorithm::Lazar.new.run({:dataset_uri => @@classification_training_dataset.uri, :subjectid => @@subjectid, :prediction_algorithm => "local_svm_classification", :local_svm_kernel => "propositionalized"}).to_s
-    lazar = OpenTox::Model::Lazar.find model_uri, @@subjectid
-    @models << lazar
-    assert_equal lazar.features.size, 52
-
-    # single prediction
-    compound = OpenTox::Compound.from_smiles("c1ccccc1NN")
-    prediction_uri = lazar.run(:compound_uri => compound.uri, :subjectid => @@subjectid)
-    prediction = OpenTox::LazarPrediction.find(prediction_uri, @@subjectid)
-    @predictions << prediction
-    assert_equal prediction.value(compound), "false"
-    assert_equal prediction.confidence(compound).round_to(4), 0.4131.round_to(4)
-    assert_equal prediction.neighbors(compound).size, 14
-
-    # dataset prediction
-    test_dataset = OpenTox::Dataset.create_from_csv_file("data/multicolumn.csv", @@subjectid)
-    prediction = OpenTox::LazarPrediction.find lazar.run(:dataset_uri => test_dataset.uri, :subjectid => @@subjectid), @@subjectid
-    @predictions << prediction
-    assert_equal prediction.compounds.size, 4
-    compound = OpenTox::Compound.from_smiles "CC(=Nc1ccc2c(c1)Cc1ccccc21)O"
-    assert_equal prediction.value(compound), nil
-    assert_equal prediction.measured_activities(compound).first, true
+    c = OpenTox::Compound.new("http://ot-dev.in-silico.ch/compound/InChI=1S/C2H4N4/c3-2-4-1-5-6-2/h1H,(H3,3,4,5,6)")
+    assert_equal 4, @predictions[1].compounds.size
+    assert_equal false, @predictions[1].value(c)
 
+    assert_equal 52, @model.features.size
+    cleanup
   end
 
 =begin
diff --git a/validate-owl.rb b/validate-owl.rb
index e219d58..2152e7d 100644
--- a/validate-owl.rb
+++ b/validate-owl.rb
@@ -2,7 +2,11 @@ def validate_owl(uri, subjectid=nil)
   if validator_available?
     owl = OpenTox::RestClientWrapper.get(uri,{:accept => "application/rdf+xml",:subjectid => subjectid}, nil, false)
     html = OpenTox::RestClientWrapper.post("http://www.mygrid.org.uk/OWL/Validator",{:rdf => owl, :level => "DL",:subjectid => subjectid})
-    assert_match(/YES/,html)
+    # assert_match(/YES/,html)
+    # avoid verbose html output if validation fails
+    owl_dl = false
+    owl_dl = true if html =~ /YES/
+    assert_equal true, owl_dl, "Invalid OWL-DL: #{uri}"
   else
     puts "http://www.mygrid.org.uk/OWL/Validator offline"
   end