production logging enabled, predictions working with documented algorithm

1 files changed, 37 insertions, 31 deletions

diff --git a/application.rb b/application.rb
index 928a122..d1aa61a 100644
--- a/application.rb
+++ b/application.rb
@@ -6,15 +6,15 @@ end
 
 get '/model/:id' do
 	halt 404, "Model #{params[:id]} not found." unless model = Model.get(params[:id])
-	halt 202, "Model #{params[:id]} still under construction, please try again later." unless model.finished
+	halt 202, model.to_yaml unless model.finished
+	model.to_yaml
 #	builder do |xml|
 #		xml.instruct!
-		model.to_yaml
 #	end
 	#xml model
 end
 
-post '/models' do # create a model
+post '/models/?' do # create a model
 
 	training_dataset = OpenTox::Dataset.new :uri => params[:dataset_uri]
 	model = Model.create(:name => training_dataset.name, :training_dataset_uri => training_dataset.uri)
@@ -59,45 +59,46 @@ post '/model/:id' do # create prediction
 
 		unless prediction.finished # present cached prediction if finished
 
-			#Spork.spork do
-				prediction.update_attributes(:uri => url_for("/prediction/", :full) + prediction.id.to_s)
+			prediction.update_attributes(:uri => url_for("/prediction/", :full) + prediction.id.to_s)
+			Spork.spork do
 				feature_dataset = OpenTox::Dataset.new :uri => model.feature_dataset_uri
-				compound_descriptors = feature_dataset.all_compounds_and_features
+				compound_descriptors = feature_dataset.all_compounds_and_features_uris
 				training_features = feature_dataset.all_features
-				compound_activities = activity_dataset.all_compounds_and_features	# TODO: returns nil/fix in gem
+				compound_activities = activity_dataset.all_compounds_and_features_uris
 				query_features = query_compound.match(training_features)
 				query_features.each do |f|
 					Feature.find_or_create(:feature_uri => f.uri, :prediction_uri => prediction.uri)
 				end
+				query_feature_uris = query_features.collect{|f| f.uri}
 
 				conf = 0.0
+				nr_neighbors = 0
 
-				compound_descriptors.each do |compound_uri,features|
-					sim = similarity(features,query_features,model)
+				compound_descriptors.each do |compound_uri,feature_uris|
+					sim = similarity(feature_uris,query_feature_uris)
 					if sim > 0.0
-						puts sim
-						Neighbor.find_or_create(:compound_uri => compound_uri, :similarity => sim, :prediction_uri => prediction.uri)
+						nr_neighbors += 1
+						# datamapper default precision is 10, floats with higher precision are not saved
+						n = Neighbor.create(:uri => compound_uri, :similarity => (1000*sim).round/1000.0, :prediction_uri => prediction.uri)
 						compound_activities[compound_uri].each do |a|
-							case a.value('classification').to_s
+							case OpenTox::Feature.new(:uri => a).value('classification').to_s
 							when 'true'
-								conf += sim #TODO gaussian
+								conf += gauss(sim) 
 							when 'false'
-								conf -= sim #TODO gaussian
+								conf -= gauss(sim)
 							end
 						end
 					end
 				end
-
+				conf = conf/nr_neighbors
 				if conf > 0.0
 					classification = true
 				elsif conf < 0.0
 					classification = false
 				end
-				prediction.update_attributes(:confidence => conf, :classification => classification, :finished => true)
-				prediction.save!
-				puts prediction.to_yaml
+				prediction.update_attributes(:confidence => (1000*conf).round/1000.0, :classification => classification, :finished => true)
 
-			#end
+			end
 			
 		end
 		
@@ -112,7 +113,6 @@ end
 
 get '/prediction/:id' do	# display prediction
 	halt 404, "Prediction #{params[:id]} not found." unless prediction = Prediction.get(params[:id])
-	#halt 202, "Prediction #{params[:id]} not yet finished, please try again later." unless prediction.finished
 	halt 202, prediction.to_yaml unless prediction.finished
 	prediction.to_yaml
 	#xml prediction
@@ -141,23 +141,29 @@ delete '/prediction/:id' do
 end
 
 # Utility functions
-def similarity(neighbor_features, query_features, model)
+def similarity(neighbor_features, query_features)
 
-	nf = neighbor_features.collect{|f| f.uri }
-	qf = query_features.collect{|f| f.uri }
-	#common_features = neighbor_features & query_features
-	#all_features    = neighbor_features | query_features
-	common_features = nf & qf
-	all_features    = nf | qf
+	common_features = neighbor_features & query_features
+	all_features    = neighbor_features | query_features
 
+	#common_features.size.to_f/all_features.size.to_f
 	sum_p_common = 0.0
 	sum_p_all = 0.0
 
-	#all_features.each { |f| sum_p_all += f.value.to_f }
-	#common_features.each { |f| sum_p_common += f.value.to_f }
-	#sum_p_common/sum_p_all
-	common_features.size.to_f/all_features.size.to_f
+	all_features.each do |f|
+		sum_p_all += gauss(OpenTox::Feature.new(:uri => f).value('p_value').to_f)
+	end
+	common_features.each do |f|
+		sum_p_common += gauss(OpenTox::Feature.new(:uri => f).value('p_value').to_f)
+	end
+	sum_p_common/sum_p_all
+
+end
 
+# gauss kernel
+def gauss(sim, sigma = 0.3) 
+	x = 1.0 - sim
+	Math.exp(-(x*x)/(2*sigma*sigma))
 end
 
 def xml(object)