Merge branch 'development' of github.com:opentox/algorithm into development

Conflicts:
	descriptor.rb
	lazar.rb
	lib/dataset.rb
	lib/lazar.rb

7 files changed, 107 insertions, 26 deletions

diff --git a/application.rb b/application.rb
index 198d3a8..1262b9e 100644
--- a/application.rb
+++ b/application.rb
@@ -17,6 +17,12 @@ Dir['./*.rb'].each { |f| require_relative f; also_reload f } # Webapps
 # Entry point
 module OpenTox
   class Application < Service
+  
+    # for service check
+    head '/?' do
+      #$logger.debug "Algorithm service is running."
+    end
+    
     get '/?' do
       list = [ to('/lazar', :full), 
                to('/fminer/bbrc', :full), 
diff --git a/descriptor.rb b/descriptor.rb
index f4597e0..23f4174 100644
--- a/descriptor.rb
+++ b/descriptor.rb
@@ -184,6 +184,50 @@ module OpenTox
           @sdf_file.close
         end
       end
+<<<<<<< HEAD
+=======
+
+      def fix_value val
+        if val.numeric?
+          val = Float(val)
+          val = nil if val.nan? or val.infinite?
+        else
+          val = nil if val == "NaN"
+        end
+        val
+      end
+    end
+
+    before '/descriptor/?*' do
+      if request.get?
+        @algorithm = OpenTox::Algorithm.new @uri, @subjectid
+        @algorithm.parameters = [ 
+          { RDF::DC.description => "Dataset URI", 
+            RDF::OT.paramScope => "optional", 
+            RDF::DC.title => "dataset_uri" } ,
+          { RDF::DC.description => "Compound URI", 
+            RDF::OT.paramScope => "optional", 
+            RDF::DC.title => "compound_uri" } 
+        ]
+        @algorithm.metadata = {
+          RDF.type => [RDF::OTA.DescriptorCalculation],
+        }
+      elsif request.post?
+        @feature_dataset = Dataset.new nil, @subjectid
+        @feature_dataset.metadata = {
+          RDF::DC.title => "Physico-chemical descriptors",
+          RDF::DC.creator => @uri,
+          RDF::OT.hasSource => @uri,
+        }
+        if params[:compound_uri]
+          @feature_dataset.parameters = [ { RDF::DC.title => "compound_uri", RDF::OT.paramValue => params[:compound_uri] }]
+        elsif params[:dataset_uri]
+          @feature_dataset.parameters = [ { RDF::DC.title => "dataset_uri", RDF::OT.paramValue => params[:dataset_uri] }]
+        else
+          bad_request_error "Please provide a dataset_uri or compound_uri parameter", @uri
+        end
+      end
+>>>>>>> ad386110267ecc3e0c5301769b4880a7e555a44e
     end
 
     before '/descriptor/:lib/:descriptor/?' do
@@ -221,7 +265,7 @@ module OpenTox
         if params[:compound_uri]
           compounds = [ Compound.new(params[:compound_uri], @subjectid) ]
         elsif params[:dataset_uri]
-          compounds = Dataset.new(params[:dataset_uri]).compounds
+          compounds = Dataset.new(params[:dataset_uri], @subjectid).compounds
         end
         [:openbabel, :cdk, :joelib].each{ |lib| send lib, compounds, descriptors[lib] if descriptors[lib] }
         @feature_dataset.put
diff --git a/fminer.rb b/fminer.rb
index 30426da..049b4e5 100644
--- a/fminer.rb
+++ b/fminer.rb
@@ -41,7 +41,7 @@ module OpenTox
     # Get representation of BBRC algorithm
     # @return [String] Representation
     get "/fminer/bbrc/?" do
-      algorithm = OpenTox::Algorithm.new(to('/fminer/bbrc',:full))
+      algorithm = OpenTox::Algorithm.new(to('/fminer/bbrc',:full), @subjectid)
       algorithm.metadata = {
         RDF::DC.title => 'Backbone Refinement Class Representatives',
         RDF::DC.creator => "andreas@maunz.de",
@@ -63,7 +63,7 @@ module OpenTox
     # Get representation of BBRC-sample algorithm
     # @return [String] Representation
     get "/fminer/bbrc/sample/?" do
-      algorithm = OpenTox::Algorithm.new(to('/fminer/bbrc/sample',:full))
+      algorithm = OpenTox::Algorithm.new(to('/fminer/bbrc/sample',:full), @subjectid)
       algorithm.metadata = {
         RDF::DC.title => 'Backbone Refinement Class Representatives, obtained from samples of a dataset',
         RDF::DC.creator => "andreas@maunz.de",
@@ -85,7 +85,7 @@ module OpenTox
     # Get representation of fminer LAST-PM algorithm
     # @return [String] Representation
     get "/fminer/last/?" do
-      algorithm = OpenTox::Algorithm.new(to('/fminer/last',:full))
+      algorithm = OpenTox::Algorithm.new(to('/fminer/last',:full), @subjectid)
       algorithm.metadata = {
         RDF::DC.title => 'Latent Structure Pattern Mining descriptors',
         RDF::DC.creator => "andreas@maunz.de",
@@ -106,7 +106,7 @@ module OpenTox
     # Get representation of matching algorithm
     # @return [String] Representation
     get "/fminer/:method/match?" do
-      algorithm = OpenTox::Algorithm.new(to("/fminer/#{params[:method]}/match",:full))
+      algorithm = OpenTox::Algorithm.new(to("/fminer/#{params[:method]}/match",:full), @subjectid)
       algorithm.metadata = {
         RDF::DC.title => 'fminer feature matching',
         RDF::DC.creator => "mguetlein@gmail.com, andreas@maunz.de",
@@ -135,7 +135,7 @@ module OpenTox
     # @return [text/uri-list] Task URI
     post '/fminer/bbrc/?' do
     
-      @@fminer=OpenTox::Algorithm::Fminer.new(to('/fminer/bbrc',:full))
+      @@fminer=OpenTox::Algorithm::Fminer.new(to('/fminer/bbrc',:full), @subjectid)
       @@fminer.check_params(params,5)
     
       task = OpenTox::Task.run("Mining BBRC features", uri('/fminer/bbrc'), @subjectid) do |task|
@@ -291,7 +291,7 @@ module OpenTox
     # @return [text/uri-list] Task URI
     post '/fminer/last/?' do
     
-      @@fminer=OpenTox::Algorithm::Fminer.new(to('/fminer/last',:full))
+      @@fminer=OpenTox::Algorithm::Fminer.new(to('/fminer/last',:full), @subjectid)
       @@fminer.check_params(params,80)
     
       task = OpenTox::Task.run("Mining LAST features", uri('/fminer/last'), @subjectid) do |task|
diff --git a/lazar.rb b/lazar.rb
index de3f753..a9ef863 100644
--- a/lazar.rb
+++ b/lazar.rb
@@ -30,8 +30,7 @@ module OpenTox
     # @return [text/uri-list] Task URI 
     post '/lazar/?' do 
       bad_request_error "Please provide a dataset_uri parameter." unless params[:dataset_uri]
-      #TODO: URI namespace clashes
-      #resource_not_found_error "Dataset '#{params[:dataset_uri]}' not found." unless URI.accessible? params[:dataset_uri]
+      resource_not_found_error "Dataset '#{params[:dataset_uri]}' not found." unless URI.accessible? params[:dataset_uri], @subjectid
       bad_request_error "Please provide a feature_generation_uri parameter." unless params[:feature_generation_uri]
       task = OpenTox::Task.run("Create lazar model", uri('/lazar'), @subjectid) do |task|
         #lazar = OpenTox::Model::Lazar.new(nil, @subjectid)
diff --git a/lib/dataset.rb b/lib/dataset.rb
new file mode 100644
index 0000000..4f51e41
--- /dev/null
+++ b/lib/dataset.rb
@@ -0,0 +1,32 @@
+=begin
+* Name: dataset.rb
+* Description: Dataset algorithms
+* Author: Andreas Maunz <andreas@maunz.de>
+* Date: 10/2012
+=end
+
+module OpenTox
+  class Dataset
+
+    # Find database activities and calculate a consens
+    # @param [Hash] uri Compound URI, Feature URI
+    # @return [Object] activity Database activity, or nil
+    def database_activity(params)
+      f=Feature.new params[:prediction_feature_uri], @subjectid
+      db_act = values(Compound.new(params[:compound_uri], @subjectid), f)
+      if !db_act.empty?
+        if f.feature_type == "classification"
+          db_act = db_act.to_scale.mode.dup
+        else
+          db_act = db_act.to_scale.median
+        end
+        $logger.debug "Database activity for '#{params[:compound_uri]}': '#{db_act}'"
+        db_act
+      else
+        nil
+      end
+    end
+
+  end
+end
+
diff --git a/lib/lazar.rb b/lib/lazar.rb
index 34d868c..edfee43 100644
--- a/lib/lazar.rb
+++ b/lib/lazar.rb
@@ -12,7 +12,6 @@ module OpenTox
     attr_accessor :prediction_dataset
 
     def initialize(params)
-      puts "prediction started"
       @prediction_dataset = OpenTox::Dataset.new(nil, @subjectid) 
       # set instance variables and prediction dataset parameters from parameters
       params.each {|k,v|
@@ -25,7 +24,7 @@ module OpenTox
         instance_variable_set("@#{k}", [])
       }
 
-      @prediction_feature = OpenTox::Feature.new(@prediction_feature_uri,@subjectid)
+      @prediction_feature = OpenTox::Feature.new @prediction_feature_uri, @subjectid
       @predicted_variable = OpenTox::Feature.new @predicted_variable_uri, @subjectid
       @predicted_confidence = OpenTox::Feature.new @predicted_confidence_uri, @subjectid
       @prediction_dataset.metadata = {
@@ -36,8 +35,6 @@ module OpenTox
         RDF::OT.predictedVariables => [@predicted_variable_uri,@predicted_confidence_uri]
       }
 
-      puts "metadata set"
-
       @training_dataset = OpenTox::Dataset.new(@training_dataset_uri,@subjectid)
 
       @feature_dataset = OpenTox::Dataset.new(@feature_dataset_uri, @subjectid)
@@ -50,18 +47,16 @@ module OpenTox
       prediction_feature_pos = @training_dataset.features.collect{|f| f.uri}.index @prediction_feature.uri
 
       if @dataset_uri
-        compounds = OpenTox::Dataset.new(@dataset_uri,@subjectid).compounds
+        compounds = OpenTox::Dataset.new(@dataset_uri, @subjectid).compounds
       else
-        compounds = [ OpenTox::Compound.new(@compound_uri,@subjectid) ]
+        compounds = [ OpenTox::Compound.new(@compound_uri, @subjectid) ]
       end
 
-      puts "compounds started"
       compounds.each do |compound|
           
 
         #database_activity = @training_dataset.database_activity(params)
         database_activities = @training_dataset.values(compound,@prediction_feature)
-        puts "checking dataset activity"
         if database_activities and !database_activities.empty?
           database_activities.each do |database_activity|
             @prediction_dataset.add_data_entry compound, @prediction_feature, database_activity
@@ -75,15 +70,10 @@ module OpenTox
             :compound => compound, 
             :feature_dataset => @feature_dataset,
           }
-          puts "fingerprints"
-          puts @feature_calculation_algorithm
-          puts compound_params
           #compound_fingerprints = OpenTox::Algorithm::FeatureValues.send( @feature_calculation_algorithm, compound_params, @subjectid )
           # TODO: fix for pc descriptors
           #compound_fingerprints = OpenTox::Algorithm::Descriptor.send( @feature_calculation_algorithm, compound, @feature_dataset.features.collect{ |f| f[RDF::DC.title] } )
           compound_fingerprints = eval("#{@feature_calculation_algorithm}(compound, @feature_dataset.features.collect{ |f| f[RDF::DC.title] } )")
-          puts "Fingerprints"
-          puts compound_fingerprints.inspect
           @training_dataset.compounds.each_with_index { |cmpd, idx|
             act = @training_dataset.data_entries[idx][prediction_feature_pos]
             @acts << (@prediction_feature.feature_type=="classification" ? @prediction_feature.value_map.invert[act] : nil)
@@ -99,12 +89,10 @@ module OpenTox
           } # query structure
 =end
           @q_prop = compound_fingerprints.first.collect{|v| v.to_f}
-          puts @q_prop.inspect
 
           mtf = OpenTox::Algorithm::Transform::ModelTransformer.new(self)
           mtf.transform
           
-          puts "Prediction"
           prediction = OpenTox::Algorithm::Neighbors.send(@prediction_algorithm, 
               { :props => mtf.props,
                 :acts => mtf.acts,
@@ -127,7 +115,7 @@ module OpenTox
       
         if @compound_uri # add neighbors only for compound predictions
           @neighbors.each do |neighbor|
-            n =  OpenTox::Compound.new(neighbor[:compound])
+            n =  OpenTox::Compound.new(neighbor[:compound], @subjectid)
             @prediction_dataset.add_data_entry n, @prediction_feature, @prediction_feature.value_map[neighbor[:activity]]
             @prediction_dataset.add_data_entry n, @similarity_feature, neighbor[:similarity]
             #@prediction_dataset << [ n, @prediction_feature.value_map[neighbor[:activity]], nil, nil, neighbor[:similarity] ]
@@ -218,6 +206,18 @@ module OpenTox
         put
         @uri
 
+<<<<<<< HEAD
+=======
+      if params[:feature_dataset_uri]
+        bad_request_error "Feature dataset #{params[:feature_dataset_uri]} does not exist." unless URI.accessible? params[:feature_dataset_uri], @subjectid
+        @parameters << {RDF::DC.title => "feature_dataset_uri", RDF::OT.paramValue => params[:feature_dataset_uri]}
+        self[RDF::OT.featureDataset] = params["feature_dataset_uri"]
+      else
+        # run feature generation algorithm
+        feature_dataset_uri = OpenTox::Algorithm.new(params[:feature_generation_uri], @subjectid).run(params)
+        @parameters << {RDF::DC.title => "feature_dataset_uri", RDF::OT.paramValue => feature_dataset_uri}
+        self[RDF::OT.featureDataset] = feature_dataset_uri
+>>>>>>> ad386110267ecc3e0c5301769b4880a7e555a44e
       end
     end
 
diff --git a/test.rb b/test.rb
index 75044da..dc5ac9f 100644
--- a/test.rb
+++ b/test.rb
@@ -6,7 +6,7 @@ module OpenTox
     post '/test/wait_for_error_in_task/?' do
       task = OpenTox::Task.run("wait_for_error_in_task",@uri,@subjectid) do |task|
         sleep 1
-        uri = OpenTox::Dataset.new(File.join($dataset[:uri],'test/error_in_task')).post
+        uri = OpenTox::Dataset.new(File.join($dataset[:uri],'test/error_in_task'), @subjectid).post
       end
       response['Content-Type'] = 'text/uri-list'
       halt 202,task.uri.to_s+"\n"