new params for new r-util funcitons

1 files changed, 7 insertions, 4 deletions

diff --git a/validation/validation_service.rb b/validation/validation_service.rb
index 1b29630..a2e97d0 100755
--- a/validation/validation_service.rb
+++ b/validation/validation_service.rb
@@ -500,7 +500,7 @@ module Validation
             meta, self.subjectid ).uri
           test_dataset_uris << test_dataset_uri
         end
-      when /^(true|super|super4)$/
+      when /^(true|super|super4|super5)$/
         raise "DO NOT USED SUPER-STRATIFICATION FOR VAL-EXPERIMENTS AND CV, IF SO SOLVE _MISSING_VAULE_NA_ PROBLEM" if stratified=~/super/
         if stratified=="true"
           features = [ self.prediction_feature ] 
@@ -627,7 +627,7 @@ module Validation
     # splits a dataset into test and training dataset
     # returns map with training_dataset_uri and test_dataset_uri
     def self.train_test_dataset_split( orig_dataset_uri, prediction_feature, subjectid, stratified="false",
-      split_ratio=nil, random_seed=nil, missing_values=nil, task=nil, features=nil )
+      split_ratio=nil, random_seed=nil, missing_values=nil, task=nil, features=nil, store_split_clusters=false )
       
       split_ratio=0.67 unless split_ratio
       split_ratio = split_ratio.to_f
@@ -651,7 +651,8 @@ module Validation
       meta = { DC.creator => $url_provider.url_for('/training_test_split',:full) }
       
       case stratified
-      when /^(true|super|super4|anti)$/
+      when /^(true|super|super4|super5|anti)$/
+        raise "store split clusters not available for true stratified splits" if store_split_clusters and stratified=="true"
         if stratified=="true"
           raise OpenTox::BadRequestError.new "prediction feature required for stratified splits" unless prediction_feature
           LOGGER.warn "split features are ignored for stratified splits (use super instead)" if features
@@ -660,10 +661,12 @@ module Validation
           LOGGER.warn "prediction feature is ignored for super- or anti-stratified splits" if prediction_feature
         end
         r_util = OpenTox::RUtil.new 
-        train, test = r_util.stratified_split( orig_dataset, meta, missing_values, split_ratio, @subjectid, random_seed, features, stratified )
+        train, test = r_util.stratified_split( orig_dataset, meta, missing_values, split_ratio, 
+          @subjectid, random_seed, features, stratified, store_split_clusters )
         r_util.quit_r
         result = {:training_dataset_uri => train.uri, :test_dataset_uri => test.uri}
       when "false"
+        raise "store split clusters not available for random splits" if store_split_clusters
         LOGGER.warn "split features are ignored for non-stratified splits (use super instead)" if features
         compounds = orig_dataset.compounds
         raise OpenTox::BadRequestError.new "Cannot split datset, num compounds in dataset < 2 ("+compounds.size.to_s+")" if compounds.size<2