diff options
Diffstat (limited to 'lib')
| -rw-r--r-- | lib/caret.rb | 11 | ||||
| -rw-r--r-- | lib/classification.rb | 5 | ||||
| -rw-r--r-- | lib/crossvalidation.rb | 1 | ||||
| -rw-r--r-- | lib/dataset.rb | 2 | ||||
| -rw-r--r-- | lib/leave-one-out-validation.rb | 1 | ||||
| -rw-r--r-- | lib/model.rb | 48 | ||||
| -rw-r--r-- | lib/regression.rb | 2 | ||||
| -rw-r--r-- | lib/train-test-validation.rb | 2 | ||||
| -rw-r--r-- | lib/unique_descriptors.rb | 3 | ||||
| -rw-r--r-- | lib/validation-statistics.rb | 8 |
10 files changed, 54 insertions, 29 deletions
diff --git a/lib/caret.rb b/lib/caret.rb index f5c2bde..8bccf74 100644 --- a/lib/caret.rb +++ b/lib/caret.rb @@ -22,12 +22,11 @@ module OpenTox end if independent_variables.flatten.uniq == ["NA"] or independent_variables.flatten.uniq == [] prediction = Algorithm::Regression::weighted_average dependent_variables:dependent_variables, weights:weights - prediction[:warning] = "No variables for regression model. Using weighted average of similar substances." + prediction[:warnings] << "No variables for regression model. Using weighted average of similar substances." elsif dependent_variables.size < 3 prediction = Algorithm::Regression::weighted_average dependent_variables:dependent_variables, weights:weights - prediction[:warning] = "Insufficient number of neighbors (#{dependent_variables.size}) for regression model. Using weighted average of similar substances." - + prediction[:warnings] << "Insufficient number of neighbors (#{dependent_variables.size}) for regression model. Using weighted average of similar substances." else dependent_variables.each_with_index do |v,i| dependent_variables[i] = to_r(v) @@ -52,7 +51,7 @@ module OpenTox $logger.debug dependent_variables $logger.debug independent_variables prediction = Algorithm::Regression::weighted_average dependent_variables:dependent_variables, weights:weights - prediction[:warning] = "R caret model creation error. Using weighted average of similar substances." + prediction[:warnings] << "R caret model creation error. Using weighted average of similar substances." return prediction end begin @@ -73,12 +72,12 @@ module OpenTox $logger.debug "R caret prediction error for:" $logger.debug self.inspect prediction = Algorithm::Regression::weighted_average dependent_variables:dependent_variables, weights:weights - prediction[:warning] = "R caret prediction error. Using weighted average of similar substances" + prediction[:warnings] << "R caret prediction error. Using weighted average of similar substances" return prediction end if prediction.nil? or prediction[:value].nil? prediction = Algorithm::Regression::weighted_average dependent_variables:dependent_variables, weights:weights - prediction[:warning] = "Could not create local caret model. Using weighted average of similar substances." + prediction[:warnings] << "Empty R caret prediction. Using weighted average of similar substances." end end prediction diff --git a/lib/classification.rb b/lib/classification.rb index 638492b..a875903 100644 --- a/lib/classification.rb +++ b/lib/classification.rb @@ -18,6 +18,11 @@ module OpenTox class_weights.each do |a,w| probabilities[a] = w.sum/weights.sum end + # DG: hack to ensure always two probability values + if probabilities.keys.uniq.size == 1 + missing_key = probabilities.keys.uniq[0].match(/^non/) ? probabilities.keys.uniq[0].sub(/non-/,"") : "non-"+probabilities.keys.uniq[0] + probabilities[missing_key] = 0.0 + end probabilities = probabilities.collect{|a,p| [a,weights.max*p]}.to_h p_max = probabilities.collect{|a,p| p}.max prediction = probabilities.key(p_max) diff --git a/lib/crossvalidation.rb b/lib/crossvalidation.rb index 75c5db5..06a1e2a 100644 --- a/lib/crossvalidation.rb +++ b/lib/crossvalidation.rb @@ -90,6 +90,7 @@ module OpenTox field :within_prediction_interval, type: Integer, default:0 field :out_of_prediction_interval, type: Integer, default:0 field :correlation_plot_id, type: BSON::ObjectId + field :warnings, type: Array end # Independent repeated crossvalidations diff --git a/lib/dataset.rb b/lib/dataset.rb index 44690e1..6e7d67f 100644 --- a/lib/dataset.rb +++ b/lib/dataset.rb @@ -46,7 +46,7 @@ module OpenTox if data_entries[substance.to_s] and data_entries[substance.to_s][feature.to_s] data_entries[substance.to_s][feature.to_s] else - nil + [nil] end end diff --git a/lib/leave-one-out-validation.rb b/lib/leave-one-out-validation.rb index 8d22018..c33c92b 100644 --- a/lib/leave-one-out-validation.rb +++ b/lib/leave-one-out-validation.rb @@ -58,6 +58,7 @@ module OpenTox field :within_prediction_interval, type: Integer, default:0 field :out_of_prediction_interval, type: Integer, default:0 field :correlation_plot_id, type: BSON::ObjectId + field :warnings, type: Array end end diff --git a/lib/model.rb b/lib/model.rb index b18610d..475a346 100644 --- a/lib/model.rb +++ b/lib/model.rb @@ -57,7 +57,7 @@ module OpenTox model.version = {:warning => "git is not installed"} end - # set defaults + # set defaults# substance_classes = training_dataset.substances.collect{|s| s.class.to_s}.uniq bad_request_error "Cannot create models for mixed substance classes '#{substance_classes.join ', '}'." unless substance_classes.size == 1 @@ -68,10 +68,6 @@ module OpenTox :method => "fingerprint", :type => "MP2D", }, - :similarity => { - :method => "Algorithm::Similarity.tanimoto", - :min => 0.1 - }, :feature_selection => nil } @@ -79,9 +75,17 @@ module OpenTox model.algorithms[:prediction] = { :method => "Algorithm::Classification.weighted_majority_vote", } + model.algorithms[:similarity] = { + :method => "Algorithm::Similarity.tanimoto", + :min => 0.1, + } elsif model.class == LazarRegression model.algorithms[:prediction] = { - :method => "Algorithm::Caret.pls", + :method => "Algorithm::Caret.rf", + } + model.algorithms[:similarity] = { + :method => "Algorithm::Similarity.tanimoto", + :min => 0.5, } end @@ -93,7 +97,7 @@ module OpenTox }, :similarity => { :method => "Algorithm::Similarity.weighted_cosine", - :min => 0.5 + :min => 0.5, }, :prediction => { :method => "Algorithm::Caret.rf", @@ -141,7 +145,6 @@ module OpenTox end model.descriptor_ids = model.fingerprints.flatten.uniq model.descriptor_ids.each do |d| - # resulting model may break BSON size limit (e.g. f Kazius dataset) model.independent_variables << model.substance_ids.collect_with_index{|s,i| model.fingerprints[i].include? d} if model.algorithms[:prediction][:method].match /Caret/ end # calculate physchem properties @@ -191,7 +194,7 @@ module OpenTox # Predict a substance (compound or nanoparticle) # @param [OpenTox::Substance] # @return [Hash] - def predict_substance substance + def predict_substance substance, threshold = self.algorithms[:similarity][:min] @independent_variables = Marshal.load $gridfs.find_one(_id: self.independent_variables_id).data case algorithms[:similarity][:method] @@ -221,20 +224,19 @@ module OpenTox bad_request_error "Unknown descriptor type '#{descriptors}' for similarity method '#{similarity[:method]}'." end - prediction = {} + prediction = {:warnings => [], :measurements => []} + prediction[:warnings] << "Similarity threshold #{threshold} < #{algorithms[:similarity][:min]}, prediction may be out of applicability domain." if threshold < algorithms[:similarity][:min] neighbor_ids = [] neighbor_similarities = [] neighbor_dependent_variables = [] neighbor_independent_variables = [] - prediction = {} # find neighbors substance_ids.each_with_index do |s,i| # handle query substance if substance.id.to_s == s - prediction[:measurements] ||= [] prediction[:measurements] << dependent_variables[i] - prediction[:warning] = "Substance '#{substance.name}, id:#{substance.id}' has been excluded from neighbors, because it is identical with the query substance." + prediction[:info] = "Substance '#{substance.name}, id:#{substance.id}' has been excluded from neighbors, because it is identical with the query substance." else if fingerprints? neighbor_descriptors = fingerprints[i] @@ -243,7 +245,7 @@ module OpenTox neighbor_descriptors = scaled_variables.collect{|v| v[i]} end sim = Algorithm.run algorithms[:similarity][:method], [similarity_descriptors, neighbor_descriptors, descriptor_weights] - if sim >= algorithms[:similarity][:min] + if sim >= threshold neighbor_ids << s neighbor_similarities << sim neighbor_dependent_variables << dependent_variables[i] @@ -258,17 +260,27 @@ module OpenTox measurements = nil if neighbor_similarities.empty? - prediction.merge!({:value => nil,:warning => "Could not find similar substances with experimental data in the training dataset.",:neighbors => []}) + prediction[:value] = nil + prediction[:warnings] << "Could not find similar substances with experimental data in the training dataset." elsif neighbor_similarities.size == 1 - prediction.merge!({:value => dependent_variables.first, :probabilities => nil, :warning => "Only one similar compound in the training set. Predicting its experimental value.", :neighbors => [{:id => neighbor_ids.first, :similarity => neighbor_similarities.first}]}) + prediction[:value] = nil + prediction[:warnings] << "Cannot create prediction: Only one similar compound in the training set." + prediction[:neighbors] = [{:id => neighbor_ids.first, :similarity => neighbor_similarities.first}] else query_descriptors.collect!{|d| d ? 1 : 0} if algorithms[:feature_selection] and algorithms[:descriptors][:method] == "fingerprint" # call prediction algorithm result = Algorithm.run algorithms[:prediction][:method], dependent_variables:neighbor_dependent_variables,independent_variables:neighbor_independent_variables ,weights:neighbor_similarities, query_variables:query_descriptors prediction.merge! result prediction[:neighbors] = neighbor_ids.collect_with_index{|id,i| {:id => id, :measurement => neighbor_dependent_variables[i], :similarity => neighbor_similarities[i]}} + #if neighbor_similarities.max < algorithms[:similarity][:warn_min] + #prediction[:warnings] << "Closest neighbor has similarity < #{algorithms[:similarity][:warn_min]}. Prediction may be out of applicability domain." + #end + end + if prediction[:warnings].empty? or threshold < algorithms[:similarity][:min] or threshold <= 0.2 + prediction + else # try again with a lower threshold + predict_substance substance, 0.2 end - prediction end # Predict a substance (compound or nanoparticle), an array of substances or a dataset @@ -300,7 +312,7 @@ module OpenTox # serialize result if object.is_a? Substance prediction = predictions[substances.first.id.to_s] - prediction[:neighbors].sort!{|a,b| b[1] <=> a[1]} # sort according to similarity + prediction[:neighbors].sort!{|a,b| b[1] <=> a[1]} if prediction[:neighbors]# sort according to similarity return prediction elsif object.is_a? Array return predictions diff --git a/lib/regression.rb b/lib/regression.rb index fd2855f..25c0732 100644 --- a/lib/regression.rb +++ b/lib/regression.rb @@ -17,7 +17,7 @@ module OpenTox sim_sum += weights[i] end if dependent_variables sim_sum == 0 ? prediction = nil : prediction = weighted_sum/sim_sum - {:value => prediction} + {:value => prediction, :warnings => ["Weighted average prediction, no prediction interval available."]} end end diff --git a/lib/train-test-validation.rb b/lib/train-test-validation.rb index 034ae3a..9a5532d 100644 --- a/lib/train-test-validation.rb +++ b/lib/train-test-validation.rb @@ -27,6 +27,8 @@ module OpenTox end end predictions.select!{|cid,p| p[:value] and p[:measurements]} + # hack to avoid mongos file size limit error on large datasets + #predictions.each{|cid,p| p[:neighbors] = []} if model.training_dataset.name.match(/mutagenicity/i) validation = self.new( :model_id => validation_model.id, :test_dataset_id => test_set.id, diff --git a/lib/unique_descriptors.rb b/lib/unique_descriptors.rb index 8341a67..fc10cd4 100644 --- a/lib/unique_descriptors.rb +++ b/lib/unique_descriptors.rb @@ -48,7 +48,8 @@ UNIQUEDESCRIPTORS = [ #"Cdk.HBondAcceptorCount", #Descriptor that calculates the number of hydrogen bond acceptors. #"Cdk.HBondDonorCount", #Descriptor that calculates the number of hydrogen bond donors. "Cdk.HybridizationRatio", #Characterizes molecular complexity in terms of carbon hybridization states. - "Cdk.IPMolecularLearning", #Descriptor that evaluates the ionization potential. + # TODO check why the next descriptor is not present in the CDK_DESCRIPTIONS variable. + #"Cdk.IPMolecularLearning", #Descriptor that evaluates the ionization potential. "Cdk.KappaShapeIndices", #Descriptor that calculates Kier and Hall kappa molecular shape indices. "Cdk.KierHallSmarts", #Counts the number of occurrences of the E-state fragments "Cdk.LargestChain", #Returns the number of atoms in the largest chain diff --git a/lib/validation-statistics.rb b/lib/validation-statistics.rb index 2d522ae..69e7992 100644 --- a/lib/validation-statistics.rb +++ b/lib/validation-statistics.rb @@ -111,6 +111,7 @@ module OpenTox # Get statistics # @return [Hash] def statistics + self.warnings = [] self.rmse = 0 self.mae = 0 self.within_prediction_interval = 0 @@ -132,8 +133,10 @@ module OpenTox end end else - warnings << "No training activities for #{Compound.find(compound_id).smiles} in training dataset #{model.training_dataset_id}." - $logger.debug "No training activities for #{Compound.find(compound_id).smiles} in training dataset #{model.training_dataset_id}." + trd_id = model.training_dataset_id + smiles = Compound.find(cid).smiles + self.warnings << "No training activities for #{smiles} in training dataset #{trd_id}." + $logger.debug "No training activities for #{smiles} in training dataset #{trd_id}." end end R.assign "measurement", x @@ -146,6 +149,7 @@ module OpenTox $logger.debug "RMSE #{rmse}" $logger.debug "MAE #{mae}" $logger.debug "#{percent_within_prediction_interval.round(2)}% of measurements within prediction interval" + $logger.debug "#{warnings}" save { :mae => mae, |