descriptor calculation for datasets working. Rakefile for compilation f Java classes.

7 files changed, 88 insertions, 52 deletions

diff --git a/descriptor.rb b/descriptor.rb
index 2d0844f..18a44d6 100644
--- a/descriptor.rb
+++ b/descriptor.rb
@@ -1,7 +1,6 @@
 # descriptors.rb
 # Calculation of physico-chemical descriptors
 # Author: Andreas Maunz, Christoph Helma
-#require 'rjb'
 require 'openbabel'
 
 module OpenTox
@@ -91,9 +90,12 @@ module OpenTox
         sdf_3d compounds
         # rjb blocks within tasks
         # Avoid "Argument list too long" error by sending only short descriptor names
-        yaml = `echo "#{@sdf}" |java -classpath #{CDK_JAR}:#{JAVA_DIR}  CdkDescriptors #{descriptors.collect{|d| d[:title].split("\s").last}.join(" ")}`
-        YAML.load(yaml).each_with_index do |calculation,i|
-          $logger.error "Descriptor calculation failed with #{$!.message} for compound #{compounds[i].uri}." if calculation.empty?
+        #yaml = `export CDKDescriptors= ;echo "#{@sdf}" |java -classpath #{CDK_JAR}:#{JAVA_DIR}  CdkDescriptors #{descriptors.collect{|d| d[:title].split("\s").last}.join(" ")}`
+        #yaml = `export CDKDescriptors='#{descriptors.collect{|d| d[:title].split("\s").last}.join(" ")}';echo "#{@sdf}" |java -classpath #{CDK_JAR}:#{JAVA_DIR}  CdkDescriptors `
+        puts `java -classpath #{CDK_JAR}:#{JAVA_DIR}  CdkDescriptors #{@sdf_file.path} #{descriptors.collect{|d| d[:title].split("\s").last}.join(" ")}`
+        #puts yaml
+        YAML.load_file(@sdf_file.path+"cdk.yaml").each_with_index do |calculation,i|
+          $logger.error "Descriptor calculation failed for compound #{compounds[i].uri}." if calculation.empty?
           calculation.each do |name,value|
             feature = DESCRIPTORS[:cdk].collect{|d| d[:features]}.flatten.select{|f| f[RDF::DC.title].split("\s").last == name.to_s}.first
             @feature_dataset.add_data_entry compounds[i], feature, fix_value(value)
@@ -104,9 +106,12 @@ module OpenTox
       def joelib compounds, descriptors
         sdf_3d compounds
         # rjb blocks within tasks
-        yaml = `echo "#{@sdf}" |java -classpath #{JOELIB_JAR}:#{JMOL_JAR}:#{LOG4J_JAR}:#{JAVA_DIR}  JoelibDescriptors #{descriptors.collect{|d| d[:java_class]}.join(" ")}|grep "^[- ]"`
-        YAML.load(yaml).each_with_index do |calculation,i|
-          $logger.error "Descriptor calculation failed with #{$!.message} for compound #{compounds[i].uri}." if calculation.empty?
+        #yaml = `echo "#{@sdf}" |java -classpath #{JOELIB_JAR}:#{JMOL_JAR}:#{LOG4J_JAR}:#{JAVA_DIR}  JoelibDescriptors #{descriptors.collect{|d| d[:java_class]}.join(" ")}|grep "^[- ]"`
+        #puts "java -classpath #{JOELIB_JAR}:#{JMOL_JAR}:#{LOG4J_JAR}:#{JAVA_DIR}  JoelibDescriptors  #{@sdf_file.path} #{descriptors.collect{|d| d[:java_class]}.join(" ")}"
+        puts `java -classpath #{JOELIB_JAR}:#{JMOL_JAR}:#{LOG4J_JAR}:#{JAVA_DIR}  JoelibDescriptors  #{@sdf_file.path} #{descriptors.collect{|d| d[:java_class]}.join(" ")}`
+        #YAML.load(yaml).each_with_index do |calculation,i|
+        YAML.load_file(@sdf_file.path+"joelib.yaml").each_with_index do |calculation,i|
+          $logger.error "Descriptor calculation failed for compound #{compounds[i].uri}." if calculation.empty?
           calculation.each do |java_class,value|
             feature = DESCRIPTORS[:joelib].select{|d| d[:java_class] == java_class}.first[:feature]
             @feature_dataset.add_data_entry compounds[i], feature, fix_value(value)
@@ -115,6 +120,7 @@ module OpenTox
       end
 
       def sdf_3d compounds
+        #unless @sdf_file and File.exists? @sdf_file.path
         unless @sdf
           @sdf = ""
           @@obconversion.set_out_format 'sdf'
@@ -125,7 +131,7 @@ module OpenTox
             OpenBabel::OBOp.find_type("Gen3D").do(@@obmol) 
             sdf_3d = @@obconversion.write_string(@@obmol)  
             if sdf_3d.match(/.nan/)
-              warning = "3D generation failed for compound #{compound.uri}, using 2D structure."
+              warning = "3D generation failed for compound #{compound.uri}, trying to calculate descriptors from 2D structure."
               $logger.warn warning
               @feature_dataset[RDF::OT.Warnings] ? @feature_dataset[RDF::OT.Warnings] << warning : @feature_dataset[RDF::OT.Warnings] = warning
               @sdf << sdf_2d
@@ -133,6 +139,9 @@ module OpenTox
               @sdf << sdf_3d
             end
           end
+          @sdf_file = Tempfile.open("sdf")
+          @sdf_file.puts @sdf
+          @sdf_file.close
         end
       end
 
@@ -184,6 +193,12 @@ module OpenTox
       @descriptor = @descriptors.first
     end
 
+    after do
+      #@sdf_file.unlink if @sdf_file and File.exists @sdf_file.path
+      #TODO cleanup yamls
+      @sdf_file = nil
+    end
+
     # Get a list of descriptor calculation 
     # @return [text/uri-list] URIs
     get '/descriptor/?' do
diff --git a/java/CdkDescriptors.class b/java/CdkDescriptors.class
index 21f82c4..9373bc7 100644
--- a/java/CdkDescriptors.class
+++ b/java/CdkDescriptors.class
diff --git a/java/CdkDescriptors.java b/java/CdkDescriptors.java
index 5635507..69c27a8 100644
--- a/java/CdkDescriptors.java
+++ b/java/CdkDescriptors.java
@@ -9,13 +9,13 @@ import org.openscience.cdk.qsar.DescriptorValue;
 class CdkDescriptors {
   public static void main(String[] args) {
 
-    // parse command line arguments (descriptors)
+    // parse command line arguments > 1 (descriptors)
     DescriptorEngine engine;
-    if (args.length > 0) {
-      for (int i =0; i < args.length; i++) {
-        args[i] = "org.openscience.cdk.qsar.descriptors.molecular." + args[i] + "Descriptor";
+    if (args.length > 1) {
+      List<String> classNames = new ArrayList<String>();
+      for (int i =1; i < args.length; i++) {
+        classNames.add("org.openscience.cdk.qsar.descriptors.molecular." + args[i] + "Descriptor");
       }
-      List<String> classNames = Arrays.asList(args);
       engine = new DescriptorEngine(classNames);
       List<IDescriptor> instances =  engine.instantiateDescriptors(classNames);
       List<DescriptorSpecification> specs = engine.initializeSpecifications(instances);
@@ -25,13 +25,21 @@ class CdkDescriptors {
       engine = new DescriptorEngine(DescriptorEngine.MOLECULAR);
     }
 
-    // parse 3d sdf from stdin and calculate descriptors
-    BufferedReader br = new BufferedReader(new InputStreamReader(System.in));
-    IteratingMDLReader reader = new IteratingMDLReader( br, DefaultChemObjectBuilder.getInstance());
-    while (reader.hasNext()) {
-      IMolecule molecule = (IMolecule)reader.next();
-      try {
-        engine.process(molecule);
+    try {
+      BufferedReader br = new BufferedReader(new FileReader(args[0]));
+      PrintWriter yaml = new PrintWriter(new FileWriter(args[0]+"cdk.yaml"));
+      // parse 3d sdf from file and calculate descriptors
+      IteratingMDLReader reader = new IteratingMDLReader( br, DefaultChemObjectBuilder.getInstance());
+      while (reader.hasNext()) {
+        try {
+          IMolecule molecule = (IMolecule)reader.next();
+          engine.process(molecule);
+        }
+        catch (Exception e) {
+          yaml.println("- {}");
+          e.printStackTrace();
+          continue;
+        }
         Iterator it = molecule.getProperties().values().iterator(); 
         Boolean first = true;
         while (it.hasNext()) {
@@ -39,28 +47,27 @@ class CdkDescriptors {
             DescriptorValue value = (DescriptorValue)it.next();
             int size = value.getValue().length();
             if (size == 1) {
-              if (first) { System.out.print("- "); }
-              else { System.out.print("  "); }
-              System.out.println(":"+value.getNames()[0].toString() + ": " + value.getValue());
+              if (first) { yaml.print("- "); }
+              else { yaml.print("  "); }
+              yaml.println(":"+value.getNames()[0].toString() + ": " + value.getValue());
               first = false;
             }
             else {
               String[] values = value.getValue().toString().split(",");
               for (int i = 0; i < size; i++) {
-                if (first) { System.out.print("- "); }
-                else { System.out.print("  "); }
-                System.out.println(":"+value.getNames()[i].toString() + ": "  + values[i]);
+                if (first) { yaml.print("- "); }
+                else { yaml.print("  "); }
+                yaml.println(":"+value.getNames()[i].toString() + ": "  + values[i]);
                 first = false;
               }
             }
           }
           catch (ClassCastException e) { } // sdf properties are stored as molecules properties (strings), ignore them
+          catch (Exception e) { e.printStackTrace(); } // output nothing to yaml
         }
       }
-      catch (Exception e) {
-        System.out.println("- {}");
-        System.err.println(e.toString());
-      }
+      yaml.close();
     }
+    catch (Exception e) { e.printStackTrace(); }
   }
 }
diff --git a/java/JoelibDescriptorInfo.class b/java/JoelibDescriptorInfo.class
index 293cb72..0ee67bf 100644
--- a/java/JoelibDescriptorInfo.class
+++ b/java/JoelibDescriptorInfo.class
diff --git a/java/JoelibDescriptors.class b/java/JoelibDescriptors.class
index 4a23d26..1426e7d 100644
--- a/java/JoelibDescriptors.class
+++ b/java/JoelibDescriptors.class
diff --git a/java/JoelibDescriptors.java b/java/JoelibDescriptors.java
index fceb2a9..ecd1b3f 100644
--- a/java/JoelibDescriptors.java
+++ b/java/JoelibDescriptors.java
@@ -15,52 +15,51 @@ import joelib2.molecule.BasicConformerMolecule;
 class JoelibDescriptors {
   public static void main(String[] args) {
 
-    // set args to all descriptors
-    if (args.length == 0) {
+    String[] features = null;
+    // set features to all descriptors
+    if (args.length == 1) {
       FeatureHelper helper = FeatureHelper.instance();
-      args = (String[]) helper.getNativeFeatures().toArray(new String[0]);
+      features = (String[]) helper.getNativeFeatures().toArray(new String[0]);
+    } else {
+      features = new String[args.length-1];
+      System.arraycopy(args,1,features,0,args.length-1);
     }
 
     FeatureFactory factory = FeatureFactory.instance();
     MoleculeFileIO loader = null;
-    BufferedReader br = new BufferedReader(new InputStreamReader(System.in));
     String line = new String();
     String sdf = new String();
     try {
-      while ((line = br.readLine()) != null) { sdf += line + "\n"; }
-      br.close();
-      InputStream is = null;
-      is = new ByteArrayInputStream(sdf.getBytes("UTF-8"));
+      // parse 3d sdf from file and calculate descriptors
+      InputStream is = new FileInputStream(args[0]);
+      PrintWriter yaml = new PrintWriter(new FileWriter(args[0]+"joelib.yaml"));
       BasicIOType inType = BasicIOTypeHolder.instance().getIOType("SDF");
       loader = MoleculeFileHelper.getMolReader(is, inType);
-      //BasicIOType outType = BasicIOTypeHolder.instance().getIOType("SMILES");
-      //JOEMol mol = new JOEMol(inType, inType);
       BasicConformerMolecule mol = new BasicConformerMolecule(inType, inType);
       while (true) {
         try {
           Boolean success = loader.read(mol);
-          if (!success) { break; }
-          //System.err.println( mol );
-          for (int i =0; i < args.length; i++) {
-            Feature feature = factory.getFeature(args[i]);
+          if (!success) { break; } // last molecule
+          for (int i =0; i < features.length; i++) {
+            Feature feature = factory.getFeature(features[i]);
             FeatureResult result = feature.calculate(mol);
-            if (i == 0) { System.out.print("- "); }
-            else { System.out.print("  "); }
-            System.out.print( args[i]+": " );
-            System.out.println( result.toString() );
+            if (i == 0) { yaml.print("- "); }
+            else { yaml.print("  "); }
+            yaml.print( features[i]+": " );
+            yaml.println( result.toString() );
           }
 
         }
         catch (Exception e) { 
-      System.err.println(e.toString());
-      e.printStackTrace();
-      //next;
+          System.err.println(e.toString());
+          e.printStackTrace();
         }
       }
+      yaml.close();
     }
     catch (Exception e) {
+      System.err.println(e.toString());
       e.printStackTrace();
-      //System.err.println(e.toString());
     }
   }
 }
diff --git a/java/Rakefile b/java/Rakefile
new file mode 100644
index 0000000..4348b49
--- /dev/null
+++ b/java/Rakefile
@@ -0,0 +1,15 @@
+# Java class, classpath
+java_classes = [
+  ["CdkDescriptors", "cdk-1.4.16.jar"],
+  ["CdkDescriptorInfo", "cdk-1.4.16.jar"],
+  ["JoelibDescriptors", "joelib2.jar:."],
+  ["JoelibDescriptorInfo", "joelib2.jar:."],
+]
+  
+task :default => java_classes.collect{|c| "#{c.first}.class"}
+
+java_classes.each do |c|
+  file "#{c.first}.class" => "#{c.first}.java" do
+    puts `javac -classpath #{c.last} #{c.first}.java`
+  end
+end