Merge remote branch 'helma/master' into development

4 files changed, 62 insertions, 225 deletions

diff --git a/Rakefile b/Rakefile
deleted file mode 100644
index f4e10d6..0000000
--- a/Rakefile
+++ /dev/null
@@ -1,20 +0,0 @@
-require 'rubygems'
-require 'rake'
-require 'tasks/opentox'
-
-desc "Install required gems"
-task :install do
-	puts `sudo gem sources -a http://gems.github.com`
-	puts `sudo gem install sinatra datamapper dm-more builder helma-opentox-ruby-api-wrapper`
-end
-
-desc "Update gems"
-task :update do
-	puts `sudo gem update sinatra datamapper dm-more builder helma-opentox-ruby-api-wrapper`
-end
-
-desc "Run tests"
-task :test do
-	load 'test/test.rb'
-end
-
diff --git a/application.rb b/application.rb
index d1ff70e..0f762eb 100644
--- a/application.rb
+++ b/application.rb
@@ -1,7 +1,6 @@
 require 'rubygems'
-gem "opentox-ruby-api-wrapper", "= 1.6.0"
+gem "opentox-ruby-api-wrapper", "= 1.6.3"
 require 'opentox-ruby-api-wrapper'
-LOGGER.progname = File.expand_path(__FILE__)
 
 class Model
 	include DataMapper::Resource
@@ -21,6 +20,15 @@ class Model
 	
 end
 
+class Prediction
+  # cache predictions
+	include DataMapper::Resource
+	property :id, Serial
+	property :compound_uri, String, :length => 255
+	property :model_uri, String, :length => 255
+	property :yaml, Text, :length => 2**32-1 
+end
+
 DataMapper.auto_upgrade!
 
 require 'lazar.rb'
@@ -68,6 +76,7 @@ end
 delete '/?' do
 	# TODO delete datasets
   Model.auto_migrate!
+  Prediction.auto_migrate!
 	response['Content-Type'] = 'text/plain'
-	"All Models deleted."
+	"All models and cached predictions deleted."
 end
diff --git a/lazar.rb b/lazar.rb
index e43995d..0fd4634 100755..100644
--- a/lazar.rb
+++ b/lazar.rb
@@ -2,8 +2,8 @@
 # workaround to initialize R non-interactively (former rinruby versions did this by default)
 # avoids compiling R with X
 R = nil
-require ("rinruby") 
-require ("haml") 
+require "rinruby" 
+require "haml" 
 
 class Lazar < Model
 
@@ -12,7 +12,7 @@ class Lazar < Model
 	# AM begin
 	# regression function, created 06/10
 	# ch: please properly integrate this into the workflow. You will need some criterium for distinguishing regression/classification (hardcoded regression for testing)
-	def regression(compound_uri,prediction)
+	def regression(compound_uri,prediction,verbose=false)
     
 		lazar = YAML.load self.yaml
 		compound = OpenTox::Compound.new(:uri => compound_uri)
@@ -21,9 +21,8 @@ class Lazar < Model
 		compound_matches = compound.match lazar.features
 
 		conf = 0.0
-		similarities = {}
-		activities = {}
-		fragments = { :activating => {}, :deactivating => {} }
+		features = { :activating => [], :deactivating => [] }
+		neighbors = {}
 		regression = nil
 
 		regr_occurrences = [] # occurrence vector with {0,1} entries
@@ -38,9 +37,19 @@ class Lazar < Model
 			sim = OpenTox::Algorithm::Similarity.weighted_tanimoto(compound_matches,matches,lazar.p_values)
 			lazar.activities[uri].each do |act|
 				if sim > 0.3
-					similarities[uri] = sim
-					activities[uri] = [] unless activities[uri]
-					activities[uri] << act
+          neighbors[uri] = {:similarity => sim}
+          neighbors[uri][:features] = { :activating => [], :deactivating => [] } unless neighbors[uri][:features]
+          matches.each do |m|
+            if lazar.effects[m] == 'activating'
+              neighbors[uri][:features][:activating] << {:smarts => m, :p_value => lazar.p_values[m]}
+            elsif lazar.effects[m] == 'deactivating'
+              neighbors[uri][:features][:deactivating] << {:smarts => m, :p_value => lazar.p_values[m]}
+            end
+          end
+          lazar.activities[uri].each do |act|
+            neighbors[uri][:activities] = [] unless neighbors[uri][:activities]
+            neighbors[uri][:activities] << act
+          end
 					conf += OpenTox::Utils.gauss(sim)
 					sims << OpenTox::Utils.gauss(sim)
 					#TODO check for 0 s
@@ -51,7 +60,7 @@ class Lazar < Model
 				end
 			end
 		end
-		conf = conf/similarities.size
+		conf = conf/neighbors.size
 		LOGGER.debug "Regression: found " + neighbor_matches.size.to_s + " neighbors."
 
 
@@ -107,32 +116,30 @@ class Lazar < Model
 			prediction.compounds << compound_uri
 			prediction.features << feature_uri 
 			prediction.data[compound_uri] = [] unless prediction.data[compound_uri]
-			compound_matches.each { |m| fragments[lazar.effects[m].to_sym][m] = lazar.p_values[m] }
+			compound_matches.each { |m| features[lazar.effects[m].to_sym] << {:smarts => m, :p_value => lazar.p_values[m] } }
 			tuple = { 
 					File.join(@@config[:services]["opentox-model"],"lazar#regression") => regression,
-					File.join(@@config[:services]["opentox-model"],"lazar#confidence") => conf,
-					#File.join(@@config[:services]["opentox-model"],"lazar#similarities") => similarities,
-					#File.join(@@config[:services]["opentox-model"],"lazar#activities") => activities,
-					#File.join(@@config[:services]["opentox-model"],"lazar#features") => fragments
+					File.join(@@config[:services]["opentox-model"],"lazar#confidence") => conf
 			}
+      if verbose
+        tuple[File.join(@@config[:services]["opentox-model"],"lazar#neighbors")] = neighbors
+        tuple[File.join(@@config[:services]["opentox-model"],"lazar#features")] = features
+      end
 			prediction.data[compound_uri] << {feature_uri => tuple}
 		end
 
-
 	end
 	# AM end
 
 
-	def classification(compound_uri,prediction)
+	def classification(compound_uri,prediction,verbose=false)
     
 		lazar = YAML.load self.yaml
 		compound = OpenTox::Compound.new(:uri => compound_uri)
 		compound_matches = compound.match lazar.features
 
 		conf = 0.0
-		similarities = {}
-		#activities = {}
-		fragments = { :activating => {}, :deactivating => {} }
+		features = { :activating => [], :deactivating => [] }
 		neighbors = {}
 		classification = nil
 
@@ -140,7 +147,6 @@ class Lazar < Model
 
 			sim = OpenTox::Algorithm::Similarity.weighted_tanimoto(compound_matches,matches,lazar.p_values)
 			if sim > 0.3
-				similarities[uri] = sim
 				neighbors[uri] = {:similarity => sim}
 				neighbors[uri][:features] = { :activating => [], :deactivating => [] } unless neighbors[uri][:features]
 				matches.each do |m|
@@ -150,11 +156,7 @@ class Lazar < Model
 						neighbors[uri][:features][:deactivating] << {:smarts => m, :p_value => lazar.p_values[m]}
 					end
 				end
-				#neighbors[uri][:features] = [] unless neighbors[uri][:features]
-				#neighbors[uri][:features] << matches
 				lazar.activities[uri].each do |act|
-					#activities[uri] = [] unless activities[uri]
-					#activities[uri] << act
 					neighbors[uri][:activities] = [] unless neighbors[uri][:activities]
 					neighbors[uri][:activities] << act
 					case act.to_s
@@ -167,7 +169,7 @@ class Lazar < Model
 			end
 		end
 	
-		conf = conf/similarities.size
+		conf = conf/neighbors.size
 		if conf > 0.0
 			classification = true
 		elsif conf < 0.0
@@ -178,17 +180,17 @@ class Lazar < Model
 			prediction.compounds << compound_uri
 			prediction.features << feature_uri 
 			prediction.data[compound_uri] = [] unless prediction.data[compound_uri]
-			compound_matches.each { |m| fragments[lazar.effects[m].to_sym][m] = lazar.p_values[m] }
-			#fragments[:activating] = fragments[:activating].sort{|a,b| b[1] <=> a[1]}
+			compound_matches.each { |m| features[lazar.effects[m].to_sym] << {:smarts => m, :p_value => lazar.p_values[m] } }
 			tuple = { 
-					File.join(@@config[:services]["opentox-model"],"lazar#classification") => classification,
-					File.join(@@config[:services]["opentox-model"],"lazar#confidence") => conf,
-					#File.join(@@config[:services]["opentox-model"],"lazar#neighbors") => neighbors,
-					#File.join(@@config[:services]["opentox-model"],"lazar#features") => fragments
+        File.join(@@config[:services]["opentox-model"],"lazar#classification") => classification,
+        File.join(@@config[:services]["opentox-model"],"lazar#confidence") => conf
 			}
+      if verbose
+        tuple[File.join(@@config[:services]["opentox-model"],"lazar#neighbors")] = neighbors
+        tuple[File.join(@@config[:services]["opentox-model"],"lazar#features")] = features
+      end
 			prediction.data[compound_uri] << {feature_uri => tuple}
 		end
-    
 	end
 
 	def database_activity?(compound_uri,prediction)
@@ -203,10 +205,6 @@ class Lazar < Model
 			prediction.data[compound_uri] = [] unless prediction.data[compound_uri]
 			db_activities.each do |act|
 				prediction.data[compound_uri] << {feature_uri => act}
-        #tuple = { 
-        #  :classification => act}
-          #:confidence => "experimental"}
-				#prediction.data[compound_uri] << {feature_uri => tuple}
 			end
 			true
 		else
@@ -220,7 +218,6 @@ class Lazar < Model
 		feature_dataset = YAML.load(RestClient.get(data.feature_dataset_uri, :accept => 'application/x-yaml').to_s)
 		owl = OpenTox::Owl.create 'Model', uri
     owl.set("creator","http://github.com/helma/opentox-model")
-		# TODO 
 		owl.set("title", URI.decode(data.dependentVariables.split(/#/).last) )
     #owl.set("title","#{URI.decode(activity_dataset.title)} lazar classification")
     owl.set("date",created_at.to_s)
@@ -321,41 +318,26 @@ post '/:id/?' do # create prediction
 	end
 
 	if compound_uri
-		# AM: switch here between regression and classification
-		begin
-			eval "lazar.#{prediction_type}(compound_uri,@prediction) unless lazar.database_activity?(compound_uri,@prediction)"
-		rescue
-			LOGGER.error "#{prediction_type} failed for #{compound_uri} with #{$!} "
-			halt 500, "Prediction of #{compound_uri} failed."
-		end
+    # look for cached prediction first
+    if cached_prediction = Prediction.first(:model_uri => lazar.uri, :compound_uri => compound_uri)
+      @prediction = YAML.load(cached_prediction.yaml)
+    else
+      begin
+        # AM: switch here between regression and classification
+        eval "lazar.#{prediction_type}(compound_uri,@prediction,true) unless lazar.database_activity?(compound_uri,@prediction)"
+        Prediction.create(:model_uri => lazar.uri, :compound_uri => compound_uri, :yaml => @prediction.to_yaml)
+      rescue
+        LOGGER.error "#{prediction_type} failed for #{compound_uri} with #{$!} "
+        halt 500, "Prediction of #{compound_uri} failed."
+      end
+    end
 		case request.env['HTTP_ACCEPT']
 		when /yaml/ 
 			@prediction.to_yaml
 		when 'application/rdf+xml'
 			@prediction.to_owl
-		when /html/
-			@compound = OpenTox::Compound.new(:uri => compound_uri)
-			@title = @prediction.title
-			if @prediction.data[@compound.uri]
-				if @prediction.creator.to_s.match(/model/) # real prediction
-					p = @prediction.data[@compound.uri].first.values.first
-					if !p[File.join(@@config[:services]["opentox-model"],"lazar#classification")].nil?
-						feature = File.join(@@config[:services]["opentox-model"],"lazar#classification")
-					elsif !p[File.join(@@config[:services]["opentox-model"],"lazar#regression")].nil?
-						feature = File.join(@@config[:services]["opentox-model"],"lazar#regression")
-					end
-					@activity = p[feature]
-					@confidence = p[File.join(@@config[:services]["opentox-model"],"lazar#confidence")]
-					@neighbors = p[File.join(@@config[:services]["opentox-model"],"lazar#neighbors")]#.sort{|a,b| b.last[:similarity] <=> a.last[:similarity]}
-					#@training_activities = p[File.join(@@config[:services]["opentox-model"],"lazar#activities")]
-					@features = p[File.join(@@config[:services]["opentox-model"],"lazar#features")]
-				else # database value
-					@measured_activities = @prediction.data[@compound.uri].first.values
-				end
-			else
-				@activity = "not available (no similar compounds in the training dataset)"
-			end
-			haml :prediction
+    else
+      halt 400, "MIME type \"#{request.env['HTTP_ACCEPT']}\" not supported." 
 		end
 	elsif dataset_uri
     response['Content-Type'] = 'text/uri-list'
@@ -379,4 +361,3 @@ post '/:id/?' do # create prediction
 	end
 
 end
-
diff --git a/views/prediction.haml b/views/prediction.haml
deleted file mode 100644
index 0187f87..0000000
--- a/views/prediction.haml
+++ /dev/null
@@ -1,133 +0,0 @@
-.predictions
-  %table
-    %tr
-      %th= @title.gsub(/_lazar_.*$/,' ').capitalize
-      %th Prediction
-      %th 
-        %a{:href => "#", :id => "linkConfidence#{p.object_id}"} Confidence
-        :javascript
-          $("a#linkConfidence#{p.object_id}").click(function () {
-            $("dl#confidence").toggle();  
-          });
-      %th Relevant features
-    %tr
-      %th
-        %img{:src => @compound.image_uri, :alt => @compound.smiles}
-        %td
-          - if @measured_activities
-            %br
-            - @measured_activities.each do |a|
-              - if activity(a) == 'active'
-                .active
-                  = activity(a)
-              - elsif activity(a) == 'inactive'
-                .inactive
-                  = activity(a)
-              - else
-                = a
-            %br
-              (
-              %a{:href => "#", :id => "linkTrainingData#{p.object_id}"} Training data
-              :javascript
-                $("a#linkTrainingData#{p.object_id}").click(function () {
-                  $("dl#training_data").toggle();  
-                });
-              )
-
-          - else
-            - if activity(@activity) == 'active'
-              .active
-                = activity(@activity)
-            - elsif activity(@activity) == 'inactive'
-              .inactive
-                = activity(@activity)
-            - elsif @activity.is_a?(Float)
-              .other
-                = sprintf('%.03g', @activity)
-            - else
-              .other
-                %em= @activity.to_s
-        %td
-          = sprintf('%.03g', @confidence.to_f.abs) if @confidence
-        %td
-          %table
-            %tr
-              %th{:colspan => 2} activating
-              %th p value
-            - if @features[:activating]
-              - @features[:activating].sort{|a,b| b.last <=> a.last }.each do |f|
-                %tr
-                  %th= f[0]
-                  %td= f[1]
-            %tr
-              %th{:colspan => 2} deactivating
-              %th p value
-            - if @features[:deactivating]
-              - @features[:deactivating].sort{|a,b| b.last <=> a.last }.each do |f|
-                %tr
-                  %th= f[0]
-                  %td= f[1]
-    %tr
-      %th Neighbors
-      %th Activity
-      %th Similarity (activity specific)
-      %th Relevant features
-    - @neighbors.sort{|a,b| b.last[:similarity] <=> a.last[:similarity]}.each do |uri,data|
-      - c = OpenTox::Compound.new(:uri => uri)
-      %tr
-        %th 
-          %br= c.smiles
-          %br
-            %a{:href => c.image_uri, :target => "_blank"} Image
-          %br
-            %img{:src => c.image_uri, :alt => c.smiles}
-        %td
-          - data[:activities].each do |act|
-            - if activity(act) == 'active'
-              .active
-                = activity(act)
-            - elsif activity(act) == 'inactive'
-              .inactive
-                = activity(act)
-            - elsif act.is_a?(Float)
-              .other
-                = sprintf('%.03g', act)
-            - else
-              .other
-                %em= act.to_s
-        %td
-          = sprintf('%.03g', data[:similarity])
-        %td
-          %table
-            %tr
-              %th{:colspan => 2} activating
-              %th p value
-              -#%td= data[:features].inspect
-            -# data[:features][:activating].each do |f|
-            - data[:features][:activating].sort{|a,b| b.last[:p_value] <=> a.last[:p_value] }.each do |f|
-              -# f.inspect
-              %tr
-                %th= f[:smarts]
-                %td= f[:p_value]
-                %td=# f[:p_value]
-            %tr
-              %th{:colspan => 2} deactivating
-              %th p value
-            -# data[:features][:deactivating].sort{|a,b| b.last[:p_value] <=> a.last[:p_value] }.each do |f|
-            - data[:features][:deactivating].each do |f|
-              %tr
-                %th= f[:smarts]
-                %td= f[:p_value]
-
-
-
-%dl#confidence{ :style => "display: none;" }
-  %dt Confidence:
-  %dd Indicates the applicability domain of a model. Predictions with a high confidence can be expected to be more reliable than predictions with low confidence. Confidence values may take any value between 0 and 1. For most models confidence &gt; 0.025 is a sensible (hard) cutoff to distiguish between reliable and unreliable predictions.
-
-
-%dl#training_data{ :style => "display: none;" }
-  %dt Training data:
-  %dd Experimental result(s) from the training dataset are displayed here. 
-
-