Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[2741] trunk/scripts/src/main/java/org/dllearner/ examples/pdb
|
From: <km...@us...> - 2011-03-31 06:34:24
|
Revision: 2741
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=2741&view=rev
Author: kmpf
Date: 2011-03-31 06:34:17 +0000 (Thu, 31 Mar 2011)
Log Message:
-----------
Modified Paths:
--------------
trunk/scripts/src/main/java/org/dllearner/examples/pdb/HelixRDFCreator.java
trunk/scripts/src/main/java/org/dllearner/examples/pdb/TrainAndTestSet.java
Modified: trunk/scripts/src/main/java/org/dllearner/examples/pdb/HelixRDFCreator.java
===================================================================
--- trunk/scripts/src/main/java/org/dllearner/examples/pdb/HelixRDFCreator.java 2011-03-29 16:09:59 UTC (rev 2740)
+++ trunk/scripts/src/main/java/org/dllearner/examples/pdb/HelixRDFCreator.java 2011-03-31 06:34:17 UTC (rev 2741)
@@ -1,6 +1,7 @@
package org.dllearner.examples.pdb;
import java.io.File;
+import java.io.FileNotFoundException;
import java.io.IOException;
import java.io.PrintStream;
@@ -11,6 +12,15 @@
import java.util.Iterator;
import java.util.Set;
+import org.dllearner.cli.Start;
+import org.dllearner.core.ComponentInitException;
+import org.dllearner.core.ComponentManager;
+import org.dllearner.core.KnowledgeSource;
+import org.dllearner.core.ReasonerComponent;
+import org.dllearner.core.owl.Description;
+import org.dllearner.core.owl.Individual;
+import org.dllearner.kb.OWLFile;
+import org.dllearner.reasoning.FastInstanceChecker;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;
@@ -61,6 +71,8 @@
private static String saveDir = "../test/pdb/";
+ private static HashMap<Resource, File> confFilePerResidue;
+ private static File confFileForAll;
/**
* @param args
@@ -70,8 +82,9 @@
TrainAndTestSet sets;
PdbRdfModel trainmodel = new PdbRdfModel();
- do{
- sets = new TrainAndTestSet(1);
+ //do{
+ String pdbID = "3LQH";
+ sets = new TrainAndTestSet(pdbID);
trainmodel.add(getRdfModelForIds(sets.getTrainset()));
/*
* String[] id = {"200L"};
@@ -82,12 +95,12 @@
/*
* as we have to handle several amino acid chains we need the first
- * amino acid of every chain, they are returned as ResIterator
+ * amino acid of every chain, they are returned within a ResIterator
*/
ResIterator niter = getFirstAA(trainmodel);
/*
- * we add some distance Information to out model
+ * we add some distance Information to our model
*/
trainmodel = addDistanceInfo(trainmodel);
@@ -96,7 +109,7 @@
* global positives ArrayList, and all others in the global negatives ArrayList
*/
createPositivesAndNegatives(niter, trainmodel);
- } while(positives.size() > 100 && negatives.size() > 100);
+ //} while(positives.size() > 100 && negatives.size() > 100);
/*
* remove all triples that contain information about begin and end of helices
@@ -108,21 +121,22 @@
Resource residue = ResourceFactory.createResource("http://bio2rdf.org/pdb:Residue");
trainmodel = removeStatementsWithObject(trainmodel, residue);
- /*
- * writes the conf-Files
- */
+ SimpleDateFormat df = new SimpleDateFormat("_yyyy_MM_dd_HH:mm");
+ String date = df.format(new Date());
+ String rdffile= "Helixtrainer" + date + ".rdf";
+ String filename = saveDir + rdffile;
-
try
{
- SimpleDateFormat df = new SimpleDateFormat("_yyyy_MM_dd_HH:mm");
- String date = df.format(new Date());
- String rdffile= "Helixtrainer" + date + ".rdf";
- String filename = saveDir + rdffile;
+ /*
+ * creatConfFile()
+ * writes the conf-Files and saves there File-objects in:
+ * confFileForAll and confFilePerResidue
+ */
createConfFile(date, rdffile, trainmodel);
PrintStream out = new PrintStream (new File(filename));
- // Output query results
+ // Output results
trainmodel.write(out, "RDF/XML");
// Important - free up resources used running the query
@@ -132,6 +146,39 @@
{
System.err.println("OutputStream konnte nicht geschlossen werden!");
}
+
+ /*
+ * load RDF file and perform learn algorithm for every .conf-file
+ */
+
+ Start start = null;
+ Iterator<Resource> aa = confFilePerResidue.keySet().iterator();
+ while ( aa.hasNext() ){
+ Resource nextRes = aa.next();
+ /*
+ ComponentManager cm = ComponentManager.getInstance();
+ KnowledgeSource ks = cm.knowledgeSource(OWLFile.class);
+ cm.applyConfigEntry(ks, "url","file://" + filename );
+ ReasonerComponent rc = cm.reasoner(FastInstanceChecker.class);
+ rc.init();
+ */
+ System.out.println(confFilePerResidue.get(nextRes).getAbsolutePath());
+ try{
+ start = new Start(confFilePerResidue.get(nextRes));
+ } catch (ComponentInitException e) {
+ e.printStackTrace();
+ } catch (FileNotFoundException e) {
+ // TODO Auto-generated catch block
+ e.printStackTrace();
+ } catch (org.dllearner.confparser.ParseException e) {
+ // TODO Auto-generated catch block
+ e.printStackTrace();
+ }
+ start.start(false);
+ Description d = start.getLearningAlgorithm().getCurrentlyBestDescription();
+ System.out.println(d.toKBSyntaxString());
+ }
+
}
private static PdbRdfModel getRdfModelForIds(String[] pdbIDs) {
@@ -239,7 +286,6 @@
" ?x2 pdb:isImmediatelyBefore ?x3 . " +
" ?x3 pdb:isImmediatelyBefore ?x4 . " +
" ?x4 pdb:isImmediatelyBefore ?x5 . }";
- System.out.println(queryString);
Query query = QueryFactory.create(queryString);
QueryExecution qe = QueryExecutionFactory.create(query, model);
model.add(qe.execConstruct());
@@ -343,40 +389,67 @@
{
// the file with all amino acids
String pdbname = saveDir + "pdb" + date + ".conf";
- PrintStream out = new PrintStream (new File(pdbname));
+ confFileForAll = new File(pdbname);
+ PrintStream out = new PrintStream (confFileForAll);
// add import statements
out.println("import(\"AA_properties.owl\");");
out.println("import(\"" + rdffile + "\");");
out.println();
+ HashMap<Resource, File> resConfFiles = new HashMap<Resource, File>(30);
+ resConfFiles.put(ala, new File(saveDir + ala.getLocalName() + date + ".conf"));
+ resConfFiles.put(cys, new File(saveDir + cys.getLocalName() + date + ".conf"));
+ resConfFiles.put(asp, new File(saveDir + asp.getLocalName() + date + ".conf"));
+ resConfFiles.put(glu, new File(saveDir + glu.getLocalName() + date + ".conf"));
+ resConfFiles.put(phe, new File(saveDir + phe.getLocalName() + date + ".conf"));
+ resConfFiles.put(gly, new File(saveDir + gly.getLocalName() + date + ".conf"));
+ resConfFiles.put(his, new File(saveDir + his.getLocalName() + date + ".conf"));
+ resConfFiles.put(ile, new File(saveDir + ile.getLocalName() + date + ".conf"));
+ resConfFiles.put(lys, new File(saveDir + lys.getLocalName() + date + ".conf"));
+ resConfFiles.put(leu, new File(saveDir + leu.getLocalName() + date + ".conf"));
+ resConfFiles.put(met, new File(saveDir + met.getLocalName() + date + ".conf"));
+ resConfFiles.put(asn, new File(saveDir + asn.getLocalName() + date + ".conf"));
+ resConfFiles.put(pro, new File(saveDir + pro.getLocalName() + date + ".conf"));
+ resConfFiles.put(gln, new File(saveDir + gln.getLocalName() + date + ".conf"));
+ resConfFiles.put(arg, new File(saveDir + arg.getLocalName() + date + ".conf"));
+ resConfFiles.put(ser, new File(saveDir + ser.getLocalName() + date + ".conf"));
+ resConfFiles.put(thr, new File(saveDir + thr.getLocalName() + date + ".conf"));
+ resConfFiles.put(val, new File(saveDir + val.getLocalName() + date + ".conf"));
+ resConfFiles.put(trp, new File(saveDir + trp.getLocalName() + date + ".conf"));
+ resConfFiles.put(tyr, new File(saveDir + tyr.getLocalName() + date + ".conf"));
+ resConfFiles.put(sel, new File(saveDir + sel.getLocalName() + date + ".conf"));
+ confFilePerResidue = resConfFiles;
+
+
+
// put all amino acid resources and the their conf-files together
HashMap<Resource, PrintStream> resprint = new HashMap<Resource, PrintStream>(30);
- resprint.put(ala, new PrintStream(new File(saveDir + ala.getLocalName() + date + ".conf")));
- resprint.put(cys, new PrintStream(new File(saveDir + cys.getLocalName() + date + ".conf")));
- resprint.put(asp, new PrintStream(new File(saveDir + asp.getLocalName() + date + ".conf")));
- resprint.put(glu, new PrintStream(new File(saveDir + glu.getLocalName() + date + ".conf")));
- resprint.put(phe, new PrintStream(new File(saveDir + phe.getLocalName() + date + ".conf")));
- resprint.put(gly, new PrintStream(new File(saveDir + gly.getLocalName() + date + ".conf")));
- resprint.put(his, new PrintStream(new File(saveDir + his.getLocalName() + date + ".conf")));
- resprint.put(ile, new PrintStream(new File(saveDir + ile.getLocalName() + date + ".conf")));
- resprint.put(lys, new PrintStream(new File(saveDir + lys.getLocalName() + date + ".conf")));
- resprint.put(leu, new PrintStream(new File(saveDir + leu.getLocalName() + date + ".conf")));
- resprint.put(met, new PrintStream(new File(saveDir + met.getLocalName() + date + ".conf")));
- resprint.put(asn, new PrintStream(new File(saveDir + asn.getLocalName() + date + ".conf")));
- resprint.put(pro, new PrintStream(new File(saveDir + pro.getLocalName() + date + ".conf")));
- resprint.put(gln, new PrintStream(new File(saveDir + gln.getLocalName() + date + ".conf")));
- resprint.put(arg, new PrintStream(new File(saveDir + arg.getLocalName() + date + ".conf")));
- resprint.put(ser, new PrintStream(new File(saveDir + ser.getLocalName() + date + ".conf")));
- resprint.put(thr, new PrintStream(new File(saveDir + thr.getLocalName() + date + ".conf")));
- resprint.put(val, new PrintStream(new File(saveDir + val.getLocalName() + date + ".conf")));
- resprint.put(trp, new PrintStream(new File(saveDir + trp.getLocalName() + date + ".conf")));
- resprint.put(tyr, new PrintStream(new File(saveDir + tyr.getLocalName() + date + ".conf")));
- resprint.put(sel, new PrintStream(new File(saveDir + sel.getLocalName() + date + ".conf")));
+ resprint.put(ala, new PrintStream(resConfFiles.get(ala)));
+ resprint.put(cys, new PrintStream(resConfFiles.get(cys)));
+ resprint.put(asp, new PrintStream(resConfFiles.get(asp)));
+ resprint.put(glu, new PrintStream(resConfFiles.get(glu)));
+ resprint.put(phe, new PrintStream(resConfFiles.get(phe)));
+ resprint.put(gly, new PrintStream(resConfFiles.get(gly)));
+ resprint.put(his, new PrintStream(resConfFiles.get(his)));
+ resprint.put(ile, new PrintStream(resConfFiles.get(ile)));
+ resprint.put(lys, new PrintStream(resConfFiles.get(lys)));
+ resprint.put(leu, new PrintStream(resConfFiles.get(leu)));
+ resprint.put(met, new PrintStream(resConfFiles.get(met)));
+ resprint.put(asn, new PrintStream(resConfFiles.get(asn)));
+ resprint.put(pro, new PrintStream(resConfFiles.get(pro)));
+ resprint.put(gln, new PrintStream(resConfFiles.get(gln)));
+ resprint.put(arg, new PrintStream(resConfFiles.get(arg)));
+ resprint.put(ser, new PrintStream(resConfFiles.get(ser)));
+ resprint.put(thr, new PrintStream(resConfFiles.get(thr)));
+ resprint.put(val, new PrintStream(resConfFiles.get(val)));
+ resprint.put(trp, new PrintStream(resConfFiles.get(trp)));
+ resprint.put(tyr, new PrintStream(resConfFiles.get(tyr)));
+ resprint.put(sel, new PrintStream(resConfFiles.get(sel)));
Property type = ResourceFactory.createProperty("http://www.w3.org/1999/02/22-rdf-syntax-ns#", "type");
// add import statements to .conf files for amino acids
- Iterator<Resource> keys = resprint.keySet().iterator();
+ Iterator<Resource> keys = resprint.keySet().iterator();
while (keys.hasNext()){
Resource k = keys.next();
Modified: trunk/scripts/src/main/java/org/dllearner/examples/pdb/TrainAndTestSet.java
===================================================================
--- trunk/scripts/src/main/java/org/dllearner/examples/pdb/TrainAndTestSet.java 2011-03-29 16:09:59 UTC (rev 2740)
+++ trunk/scripts/src/main/java/org/dllearner/examples/pdb/TrainAndTestSet.java 2011-03-31 06:34:17 UTC (rev 2741)
@@ -23,6 +23,15 @@
public String[] getTestset() {
return testset;
}
+
+ public TrainAndTestSet (String pdbID) {
+ String[] pdbIDs = {pdbID};
+ this.trainset = pdbIDs;
+ }
+
+ public TrainAndTestSet (String[] pdbIDs) {
+ this.trainset = pdbIDs;
+ }
public TrainAndTestSet (int setsize) {
@@ -50,8 +59,8 @@
}
// lets create Train- and Testset
- this.trainset = this.create_set(setsize, linenr);
- this.testset = this.create_set(setsize, linenr);
+ this.trainset = this.createSet(setsize, linenr);
+ this.testset = this.createSet(setsize, linenr);
} catch (MalformedURLException e) {
// TODO Auto-generated catch block
@@ -108,7 +117,7 @@
*/
//creates Sets of PDB IDs equal to setsize
- private String [] create_set(int setsize, int linenr){
+ private String [] createSet(int setsize, int linenr){
String [] set = new String [setsize];
if (this.setentries == null) {
this.setentries = new HashMap<Integer,String>(2*setsize);
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
