Finished basic integration of LAST-PM webservice

TODOs:

- implement (de)activating facility in ruby (ch has still the bug for BBRCs in his code)
- implement p-value facility in ruby
- solve library 'require' issue: need completely separate namespaces or completely same ns and try to re-require after deleting from $''.
- Merge last-utils 'experimental' to 'master'

4 files changed, 113 insertions, 131 deletions

diff --git a/application.rb b/application.rb
index 3d511d1..5aecfbb 100644
--- a/application.rb
+++ b/application.rb
@@ -1,6 +1,4 @@
 require 'rubygems'
-require File.join(File.expand_path(File.dirname(__FILE__)), 'libfminer/libbbrc/bbrc') # has to be included before openbabel, otherwise we have strange SWIG overloading problems
-require File.join(File.expand_path(File.dirname(__FILE__)), 'libfminer/liblast/last') # has to be included before openbabel, otherwise we have strange SWIG overloading problems
 require File.join(File.expand_path(File.dirname(__FILE__)), 'last-utils/lu.rb') # AM LAST
 gem "opentox-ruby-api-wrapper", "= 1.6.5"
 require 'opentox-ruby-api-wrapper'
diff --git a/fminer.rb b/fminer.rb
index 6ba2834..8cbf490 100644
--- a/fminer.rb
+++ b/fminer.rb
@@ -12,13 +12,14 @@ get '/fminer/?' do
   }
   rdf = owl.rdf
   File.open('public/fminer.owl', 'w') {|f| f.print rdf}
-	response['Content-Type'] = 'application/rdf+xml'
-	rdf
+  response['Content-Type'] = 'application/rdf+xml'
+  rdf
 end
 
 
 ['/fminer/bbrc/?','/fminer/?'].each do |path| # AM LAST: set bbrc as default
   post path do
+    require File.join(File.expand_path(File.dirname(__FILE__)), 'libfminer/libbbrc/bbrc') # has to be included before openbabel, otherwise we have strange SWIG overloading problems
     @@fminer = Bbrc::Bbrc.new 
     halt 404, "Please submit a dataset_uri." unless params[:dataset_uri] and  !params[:dataset_uri].nil?
     halt 404, "Please submit a feature_uri." unless params[:feature_uri] and  !params[:feature_uri].nil?
@@ -50,7 +51,7 @@ end
       @@fminer.Reset
       #@@fminer.SetChisqSig(0.99)
       LOGGER.debug "Fminer: initialising ..."
-          training_dataset.data.each do |c,features|
+      training_dataset.data.each do |c,features|
         begin
           smiles = OpenTox::Compound.new(:uri => c.to_s).smiles
         rescue
@@ -119,17 +120,17 @@ end
             end
           else #regression part
             ids = f[2]
-                      # DV: effect calculation
-                      f_arr=Array.new
-                      f[2].each do |id|
-                          f_arr.push(g_hash[id]) 
-                      end 
-                      f_median=OpenTox::Utils.median(f_arr)
-                      if g_median >= f_median 
+            # DV: effect calculation
+            f_arr=Array.new
+            f[2].each do |id|
+              f_arr.push(g_hash[id]) 
+            end 
+            f_median=OpenTox::Utils.median(f_arr)
+            if g_median >= f_median 
               effect = 'activating'
             else
               effect = 'deactivating'
-                      end
+            end
           end
 
           tuple = {
@@ -157,137 +158,120 @@ end
 
 
 post '/fminer/last/?' do
+  require File.join(File.expand_path(File.dirname(__FILE__)), 'libfminer/liblast/last') # has to be included before openbabel, otherwise we have strange SWIG overloading problems
   @@fminer = Last::Last.new 
-	halt 404, "Please submit a dataset_uri." unless params[:dataset_uri] and  !params[:dataset_uri].nil?
-	halt 404, "Please submit a feature_uri." unless params[:feature_uri] and  !params[:feature_uri].nil?
-	LOGGER.debug "Dataset: " + params[:dataset_uri]
-	LOGGER.debug "Endpoint: " + params[:feature_uri]
-	feature_uri = params[:feature_uri]
-	begin
-		LOGGER.debug "Retrieving #{params[:dataset_uri]}"
-		training_dataset = OpenTox::Dataset.find "#{params[:dataset_uri]}"
-	rescue
-		LOGGER.error "Dataset #{params[:dataset_uri]} not found" 
-		halt 404, "Dataset #{params[:dataset_uri]} not found." if training_dataset.nil? 
-	end
-	halt 404, "No feature #{params[:feature_uri]} in dataset #{params[:dataset_uri]}" unless training_dataset.features and training_dataset.features.include?(params[:feature_uri])
+  halt 404, "Please submit a dataset_uri." unless params[:dataset_uri] and  !params[:dataset_uri].nil?
+  halt 404, "Please submit a feature_uri." unless params[:feature_uri] and  !params[:feature_uri].nil?
+  LOGGER.debug "Dataset: " + params[:dataset_uri]
+  LOGGER.debug "Endpoint: " + params[:feature_uri]
+  feature_uri = params[:feature_uri]
+  begin
+    LOGGER.debug "Retrieving #{params[:dataset_uri]}"
+    training_dataset = OpenTox::Dataset.find "#{params[:dataset_uri]}"
+  rescue
+    LOGGER.error "Dataset #{params[:dataset_uri]} not found" 
+    halt 404, "Dataset #{params[:dataset_uri]} not found." if training_dataset.nil? 
+  end
+  halt 404, "No feature #{params[:feature_uri]} in dataset #{params[:dataset_uri]}" unless training_dataset.features and training_dataset.features.include?(params[:feature_uri])
 
   task_uri = OpenTox::Task.as_task("Mine features", url_for('/fminer/last',:full)) do 
 
-		feature_dataset = OpenTox::Dataset.new
-		title = "LAST-PM descriptors for " + training_dataset.title
-		feature_dataset.title = title
-		feature_dataset.creator = url_for('/fminer/last',:full)
-		last_uri = url_for("/fminer#LAST-PM_descriptors",:full)
-		feature_dataset.features << last_uri
+    feature_dataset = OpenTox::Dataset.new
+    title = "LAST-PM descriptors for " + training_dataset.title
+    feature_dataset.title = title
+    feature_dataset.creator = url_for('/fminer/last',:full)
+    last_uri = url_for("/fminer#LAST-PM_descriptors",:full)
+    feature_dataset.features << last_uri
 
-		id = 1 # fminer start id is not 0
-		compounds = []
+    id = 1 # fminer start id is not 0
+    compounds = []
+    smi = [] # AM LAST: needed for matching the patterns back
 
     g_hash = Hash.new# DV: for effect calculation in regression part
-		@@fminer.Reset
+    @@fminer.Reset
     #@@fminer.SetChisqSig(0.99)
-		LOGGER.debug "Fminer: initialising ..."
-        training_dataset.data.each do |c,features|
-			begin
-				smiles = OpenTox::Compound.new(:uri => c.to_s).smiles
-			rescue
-				LOGGER.warn "No resource for #{c.to_s}"
-				next
-			end
-			if smiles == '' or smiles.nil?
-				LOGGER.warn "Cannot find smiles for #{c.to_s}."
-			else
-				feature_dataset.compounds << c.to_s
-				features.each do |feature|
-					act = feature[feature_uri]
-					if act.nil? 
-						LOGGER.warn "No #{feature_uri} activiity for #{c.to_s}."
-					else
-						case act.to_s
-						when "true"
-							#LOGGER.debug id.to_s + ' "' + smiles +'"' +  "\t" + true.to_s
-							activity = 1
-						when "false"
-							#LOGGER.debug id.to_s + ' "' + smiles +'"' +  "\t" + false.to_s
-							activity = 0
-						else
-							# AM: add quantitative activity
-							activity = act.to_f
-							@@fminer.SetRegression(true)
-						end
-						compounds[id] = c.to_s
-						begin
-							@@fminer.AddCompound(smiles,id)
-							@@fminer.AddActivity(activity, id)
+    LOGGER.debug "Fminer: initialising ..."
+    training_dataset.data.each do |c,features|
+      begin
+        smiles = OpenTox::Compound.new(:uri => c.to_s).smiles
+      rescue
+        LOGGER.warn "No resource for #{c.to_s}"
+        next
+      end
+      if smiles == '' or smiles.nil?
+        LOGGER.warn "Cannot find smiles for #{c.to_s}."
+      else
+        feature_dataset.compounds << c.to_s
+        features.each do |feature|
+          act = feature[feature_uri]
+          if act.nil? 
+            LOGGER.warn "No #{feature_uri} activiity for #{c.to_s}."
+          else
+            case act.to_s
+            when "true"
+              #LOGGER.debug id.to_s + ' "' + smiles +'"' +  "\t" + true.to_s
+              activity = 1
+            when "false"
+              #LOGGER.debug id.to_s + ' "' + smiles +'"' +  "\t" + false.to_s
+              activity = 0
+            else
+              # AM: add quantitative activity
+              activity = act.to_f
+              @@fminer.SetRegression(true)
+            end
+            compounds[id] = c.to_s 
+            smi[id] = smiles # AM LAST: changed this to store SMILES.
+            begin
+              @@fminer.AddCompound(smiles,id)
+              @@fminer.AddActivity(activity, id)
               g_hash[id]=activity # DV: insert global information
-						rescue
-							LOGGER.warn "Could not add " + smiles + "\t" + act.to_s + " to fminer"
-						end
-					end
-				end
-				id += 1
-			end
-		end
+            rescue
+              LOGGER.warn "Could not add " + smiles + "\t" + act.to_s + " to fminer"
+            end
+          end
+        end
+        id += 1
+      end
+    end
     g_array=g_hash.values # DV: calculation of global median for effect calculation
     g_median=OpenTox::Utils.median(g_array)
-		minfreq = (0.02*id).round
-		#minfreq = 5
-		@@fminer.SetMinfreq(minfreq)
-		LOGGER.debug "Fminer: initialised with #{id} compounds, minimum frequency #{minfreq}"
-
+    minfreq = (0.02*id).round
+    #minfreq = 5
+    @@fminer.SetMinfreq(minfreq)
+    LOGGER.debug "Fminer: initialised with #{id} compounds, minimum frequency #{minfreq}"
     raise "no compounds" if compounds.size==0
+    # run @@fminer
+    LOGGER.debug "Fminer: mining ..."
+    @@fminer.SetConsoleOut(false)
+    xml = ""
+    (0 .. @@fminer.GetNoRootNodes()-1).each do |j|
+      results = @@fminer.MineRoot(j)
+      results.each do |result|
+        xml << result
+      end
+    end
+    lu = LU.new
+    dom=lu.read(xml)                        # AM LAST: parse GraphML (needs hpricot, @ch: to be included in wrapper!)
+    smarts=lu.smarts_rb(dom,'msa')          # AM LAST: converts patterns to LAST-SMARTS using msa variant (see last-pm.maunz.de)
+    instances=lu.match_rb(smi,smarts)       # AM LAST: creates instantiations
 
-		values = {}
-		# run @@fminer
-		LOGGER.debug "Fminer: mining ..."
-		(0 .. @@fminer.GetNoRootNodes()-1).each do |j|
-			results = @@fminer.MineRoot(j)
-			results.each do |result|
-				f = YAML.load(result)[0]
-				smarts = f[0]
-				p_value = f[1]
-				# AM: f[3] missing on regression
-				if (!@@fminer.GetRegression) 
-					ids = f[2] + f[3]
-					if f[2].size > f[3].size
-						effect = 'activating'
-					else
-						effect = 'deactivating'
-					end
-				else #regression part
-					ids = f[2]
-                    # DV: effect calculation
-                    f_arr=Array.new
-                    f[2].each do |id|
-                        f_arr.push(g_hash[id]) 
-                    end 
-                    f_median=OpenTox::Utils.median(f_arr)
-                    if g_median >= f_median 
-						effect = 'activating'
-					else
-						effect = 'deactivating'
-                    end
-				end
-
-				tuple = {
-					url_for('/fminer#smarts',:full) => smarts,
-					url_for('/fminer#p_value',:full) => p_value.to_f,
-					url_for('/fminer#effect',:full) => effect
-				}
-				#LOGGER.debug "#{f[0]}\t#{f[1]}\t#{effect}"
-				ids.each do |id|
-					feature_dataset.data[compounds[id]] = [] unless feature_dataset.data[compounds[id]]
-					feature_dataset.data[compounds[id]] << {last_uri => tuple}
-				end
-			end
-		end
+    instances.each do |smarts, ids|
+      tuple = {
+        url_for('/fminer#smarts',:full) => smarts,
+        url_for('/fminer#p_value',:full) => nil, # AM LAST: TODO
+        url_for('/fminer#effect',:full) => nil   # AM LAST: TODO
+      }
+      ids.each do |id|
+        feature_dataset.data[compounds[id]] = [] unless feature_dataset.data[compounds[id]]
+        feature_dataset.data[compounds[id]] << {last_uri => tuple}
+      end
+    end
 
-		uri = feature_dataset.save 
-		LOGGER.debug "Fminer finished, dataset #{uri} created."
+    uri = feature_dataset.save 
+    LOGGER.debug "Fminer finished, dataset #{uri} created."
     uri
-	end
-	LOGGER.debug "Fimer task started: "+task_uri.to_s
-	response['Content-Type'] = 'text/uri-list'
-	halt 202,task_uri.to_s+"\n"
+  end
+  LOGGER.debug "Fimer task started: "+task_uri.to_s
+  response['Content-Type'] = 'text/uri-list'
+  halt 202,task_uri.to_s+"\n"
 end
diff --git a/last-utils b/last-utils
-Subproject a678d63a5ccc7a1b5375406a3da66f8bd325141
+Subproject d2ad4f2bb82fdb5433d3f739400244ba89f0786
diff --git a/libfminer b/libfminer
-Subproject e955cc6b24d577d7187e5660716ee69d12174a8
+Subproject 00ce7e88f4f1ca16030a9b7bdabf216b5402f9c