From 8adc2faf1cea832e91e08577e25831fd656120b4 Mon Sep 17 00:00:00 2001
From: mguetlein <martin.guetlein@gmail.com>
Date: Wed, 13 Jun 2012 11:25:20 +0200
Subject: make fminer matching less memmory consuming

---
 algorithm_test.rb | 31 ++++++++++++++++++++-----------
 fminer.rb         | 31 +++++++++++++++++++------------
 2 files changed, 39 insertions(+), 23 deletions(-)

diff --git a/algorithm_test.rb b/algorithm_test.rb
index 4b6da4e..428ed2b 100644
--- a/algorithm_test.rb
+++ b/algorithm_test.rb
@@ -56,18 +56,22 @@ class AlgorithmTest < Test::Unit::TestCase
       #kazius 250 no features
       dataset_uri = "http://local-ot/dataset/9264"
       prediction_feature = dataset_uri+"/feature/endpoint"
+      feature_dataset_uri = "http://local-ot/dataset/91409"
       
       params = {:dataset_uri=>dataset_uri,
                 :prediction_feature=>prediction_feature,
                 :min_frequency=>7, :max_num_features=>300} #multi: 10=>4, 5=>>3000
-      post "/fminer/bbrc",params
+      
       
 #      params = {:dataset_uri=>dataset_uri,
 #        :prediction_feature=>prediction_feature, :feature_dataset_uri=>feature_dataset_uri}
 #      post "/lazar",params
       
-      uri = wait_for_task(last_response.body)
-      puts uri
+      #post "/fminer/bbrc",params
+      #uri = wait_for_task(last_response.body)
+      #puts uri
+      
+      #puts OpenTox::RestClientWrapper.post(File.join(CONFIG[:services]["opentox-algorithm"],"fminer/bbrc"),params)
       
 #      model = uri
 #      puts "model #{model}"
@@ -76,13 +80,18 @@ class AlgorithmTest < Test::Unit::TestCase
       
       #puts "features: "+OpenTox::Dataset.find(uri).features.size.to_s
      
-     
-#      params = {:dataset_uri=>"http://local-ot/dataset/1724",
-#        :feature_dataset_uri=>"http://local-ot/dataset/1725"}
-#      #post "/fminer/match",params
-#      #uri = wait_for_task(last_response.body)
-#      #puts uri
-#      
+      feature_dataset_uri="http://opentox.informatik.uni-freiburg.de/dataset/3277"
+      dataset_uri="http://opentox.informatik.uni-freiburg.de/dataset/1333"
+
+      params = {:dataset_uri=>dataset_uri,
+        :feature_dataset_uri=>feature_dataset_uri}
+      #post "/fminer/bbrc/match",params
+      #uri = wait_for_task(last_response.body)
+      
+      puts OpenTox::RestClientWrapper.post(File.join(CONFIG[:services]["opentox-algorithm"],"fminer/bbrc/match"),params)
+      
+#      puts uri
+      
 #      fminer = File.join(CONFIG[:services]["opentox-algorithm"],"fminer/bbrc")
 #      OpenTox::RestClientWrapper.post(fminer,params)
       
@@ -92,7 +101,7 @@ class AlgorithmTest < Test::Unit::TestCase
 #      post "/lazar",params
 #      uri = wait_for_task(last_response.body)
 #      puts uri
-      puts "features: "+OpenTox::Dataset.find(uri).features.size.to_s
+      #puts "features: "+OpenTox::Dataset.find(uri).features.size.to_s
       
    #  fminer = File.join(CONFIG[:services]["opentox-algorithm"],"fminer/bbrc")
    #  OpenTox::RestClientWrapper.post(fminer,params)
diff --git a/fminer.rb b/fminer.rb
index 5c0fdf2..762b3bd 100644
--- a/fminer.rb
+++ b/fminer.rb
@@ -102,10 +102,9 @@ post '/fminer/:method/match?' do
       res_dataset.add_feature(f,m)
     end
 
-    step_width = 100 / c_dataset.compounds.size.to_f
-    count = 0 
-    
     if params[:nr_hits] == "true"
+      step_width = 100 / c_dataset.compounds.size.to_f
+      count = 0
       c_dataset.compounds.each do |c|
         res_dataset.add_compound(c)
         comp = OpenTox::Compound.new(c)
@@ -118,21 +117,29 @@ post '/fminer/:method/match?' do
       end
     else
       LOGGER.debug "match #{c_dataset.compounds.size} compounds with #{f_dataset.features.keys.size} features"
-      
+      step_width = 100 / f_dataset.features.size.to_f
+      count = 0
+        
       obconversion = OpenBabel::OBConversion.new
-      obmol = OpenBabel::OBMol.new
-      obconversion.set_in_format('inchi')
+      obconversion.set_in_format('inchi')    
       smarts_pattern = OpenBabel::OBSmartsPattern.new
+      
+      obmols = {}
       c_dataset.compounds.each do |c|
         res_dataset.add_compound(c)
-        comp = OpenTox::Compound.new(c)
-        obconversion.read_string(obmol,comp.inchi)
-        f_dataset.features.each do |f,m|
-          smarts_pattern.init(m[OT.smarts])
-          res_dataset.add(c,f,1) if smarts_pattern.match(obmol)
+        inchi = OpenTox::Compound.new(c).inchi
+        obmol = OpenBabel::OBMol.new
+        obconversion.read_string(obmol,inchi)
+        obmols[c] = obmol
+      end
+      
+      f_dataset.features.each do |f,m|
+        smarts_pattern.init(m[OT.smarts])
+        c_dataset.compounds.each do |c|
+          res_dataset.add(c,f,1) if smarts_pattern.match(obmols[c])
         end
         count += 1
-        task.progress step_width*count if count%100==0
+        task.progress step_width*count if count%10==0
       end
     end
     res_dataset.save @subjectid
-- 
cgit v1.2.3