First parser for Macot file, debugging in progress

git-svn-id: https://subversion.renater.fr/xtandempipeline/trunk@191 b8ef2a07-7df7-436f-90b9-41648038564b

xtandempipeline/.classpath

@@ -32,5 +32,16 @@
 	<classpathentry kind="lib" path="lib/swt/swt-gtk-linux-x86_64.jar"/>
 	<classpathentry kind="lib" path="lib/gnumail-1.1.2.jar"/>
 	<classpathentry kind="lib" path="lib/inetlib-1.1.1.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/mascotdatfile-3.2.9.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/lib/avalon-framework-4.1.3.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/lib/commons-io-1.3.1.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/lib/cpdetector-1.0.4.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/lib/iText-2.0.8.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/lib/junit-3.8.1.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/lib/lmjunit-1.0.2.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/lib/logkit-1.0.1.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/lib/servlet-api-2.2.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/lib/utilities-3.0.7.jar"/>
+	<classpathentry kind="lib" path="/tmp/mascotdatfile-3.2.9/lib/xxindex-0.5.jar"/>
 	<classpathentry kind="output" path="bin"/>
 </classpath>

xtandempipeline/src/fr/inra/pappso/xtandempipeline/sax_parse/Mascot.java

+package fr.inra.pappso.xtandempipeline.sax_parse;
+
+import java.util.Vector;
+
+import org.apache.log4j.Logger;
+
+import com.compomics.mascotdatfile.util.interfaces.Modification;
+import com.compomics.mascotdatfile.util.mascot.MascotDatfile;
+import com.compomics.mascotdatfile.util.mascot.PeptideHit;
+import com.compomics.mascotdatfile.util.mascot.ProteinHit;
+import com.compomics.mascotdatfile.util.mascot.Query;
+import com.compomics.mascotdatfile.util.mascot.QueryToPeptideMap;
+
+import fr.inra.pappso.xtandempipeline.class_msms.Identification;
+import fr.inra.pappso.xtandempipeline.class_msms.Match;
+import fr.inra.pappso.xtandempipeline.class_msms.Modifs;
+import fr.inra.pappso.xtandempipeline.class_msms.Peptide;
+import fr.inra.pappso.xtandempipeline.class_msms.Protein;
+import fr.inra.pappso.xtandempipeline.class_msms.IdentificationDataFile.IdentificationDataFile;
+
+public class Mascot {
+	private static final Logger logger = Logger.getLogger(Mascot.class);
+
+	private IdentificationDataFile sample_;
+	private Identification identification_;
+
+	private MascotDatfile mascotDatfile;
+
+	private Peptide pep_;
+
+	private Protein prot_;
+
+	private Match match_;
+
+	public Mascot(Identification L, IdentificationDataFile sample) {
+		// Identification is null if want a new one!
+		setIdentification(L);
+		this.sample_ = sample;
+	}
+
+	public IdentificationDataFile getSample() {
+		return sample_;
+	}
+
+	public Identification getIdentification() {
+		return identification_;
+	}
+
+	private void setIdentification(Identification identification_) {
+		if (identification_ == null)
+			this.identification_ = new Identification();
+		else
+			this.identification_ = identification_;
+	}
+
+	public void readAndParseMascotFile() throws Exception {
+		// Load Mascot file
+		logger.info("Read :" + sample_.getName());
+		mascotDatfile = new MascotDatfile(sample_.getAbsolutePath());
+
+		// Get liste of query and associated peptide
+		QueryToPeptideMap map = mascotDatfile.getQueryToPeptideMap();
+
+		for (Object q : mascotDatfile.getQueryList()) {
+			Query query = (Query) q;
+			Vector<?> peptides = map.getAllPeptideHits(query.getQueryNumber());
+			if (peptides != null) {
+				// Liste of peptideHit
+				for (Object ob : peptides) {
+					PeptideHit pep = (PeptideHit) ob;
+
+					// Mapping peptideHit to peptide
+					pep_ = new Peptide();
+
+					pep_.set_sample_id(identification_.getMsRunSet()
+							.getInstance(sample_.getName()), sample_);
+
+					pep_.set_charge((int) round(
+							(pep.getPeptideMr() / query.getPrecursorMZ()), 0));
+					pep_.set_deltamass((float) pep.getDeltaMass());
+					pep_.set_evalue((float) pep.getExpectancy());
+					pep_.set_hypercorr((float) pep.getIonsScore());
+					pep_.set_mhplus_obser((float) (pep.getPeptideMr()
+							+ pep.getDeltaMass() + 1.007825));
+					pep_.set_mhplus_theo((float) (pep.getPeptideMr() + 1.007825));
+					pep_.set_sequence(pep.getSequence());
+
+					// Add first query
+					pep_.set_scan(query.getQueryNumber());
+					// If scan exist, add it
+					try {
+						int scan = Integer.parseInt(query.getScans());
+						pep_.set_scan(scan);
+					} catch (Exception e) {
+						logger.debug("No scan for query :"
+								+ query.getQueryNumber());
+					}
+					String rt = query.getRetentionTimeInSeconds();
+					if (rt != null)
+						pep_.set_RT(rt);
+
+					// modifs
+					Modification[] mods = pep.getModifications();
+					for (int i = 0; i < mods.length; i++) {
+						Modification mod = mods[i];
+						if (mod != null) {
+							pep_.set_Modifs(mod.getLocation(), i,
+									(float) mod.getMass());
+							identification_.add_modifs_to_liste(new Modifs(
+									(float) mod.getMass()));
+						}
+					}
+
+					// mapping proteinHit to protein
+					for (Object obj : pep.getProteinHits()) {
+						ProteinHit prot = (ProteinHit) obj;
+
+						prot_ = new Protein();
+						match_ = new Match();
+						prot_.set_description(prot.getAccession()
+								+ " "
+								+ mascotDatfile.getProteinMap()
+										.getProteinDescription(
+												prot.getAccession()));
+						prot_.set_sequence("");
+						prot_.setDatabase(identification_.getDatabaseSet()
+								.getInstance("Inconnue"));
+
+						// Get existing match or create a new one
+						if (identification_.contain_protein(prot_)) {
+							match_.set_protein_match(prot_);
+							identification_.add_protein(prot_, match_);
+						}
+
+						else {
+							match_ = (Match) identification_.get_match(prot_);
+							prot_ = match_.get_protein_match();
+						}
+
+						// Add specific peptide param to protein
+						pep_.set_post("");
+						pep_.set_pre("");
+						pep_.set_start(prot.getStart());
+						pep_.set_stop(prot.getStop());
+
+						// Add result to identification
+						identification_.add_peptide(pep_);
+						match_.add_peptide_match(pep_);
+						prot_.add_one_peptide();
+
+					}
+				}
+			}
+		}
+
+	}
+
+	public double round(double what, int howmuch) {
+		return (double) ((int) (what * Math.pow(10, howmuch) + .5))
+				/ Math.pow(10, howmuch);
+	}
+}

xtandempipeline/src/fr/inra/pappso/xtandempipeline/sax_parse/MassChroQML.java

@@ -298,7 +298,7 @@ public class MassChroQML {
 			for (int z = 0; z < mods.size(); z++) {
 				label = label + " " + "iso" + (z + 1);
 			}
-			peptides_in_peptide_list.setAttribute("isotope_label_refs", label);
+			peptides_in_peptide_list.setAttribute("isotope_label_refs", label.trim());
 		}
 		quantify.appendChild(peptides_in_peptide_list);
 		quantification.appendChild(quantify);
@@ -330,8 +330,7 @@ public class MassChroQML {
 		for(MsRun sample : ident.get_samples().getMsRunList()){
 			raw_list = raw_list + " " + sample_to_id.get(sample.getSampleName());
 		}
-		raw_list.replaceFirst(" ", "");
-		group.setAttribute("data_ids", raw_list);
+		group.setAttribute("data_ids", raw_list.trim());
 		groups.appendChild(group);
 	}
 
@@ -408,7 +407,7 @@ public class MassChroQML {
 				String ids = prots.next();
 				while (prots.hasNext())
 					ids = ids + " " + prots.next();
-				peptide.setAttribute("prot_ids", ids);
+				peptide.setAttribute("prot_ids", ids.trim());
 			}
 
 		}

xtandempipeline/src/fr/inra/pappso/xtandempipeline/sax_parse/xtandem_params.java

@@ -20,9 +20,10 @@ public class xtandem_params {
 	private InputStream template = this.getClass().getResourceAsStream(
 			"/resources/xtandem_model.xml");
 
-	public xtandem_params() {
+	public xtandem_params() throws Exception {
 		this.params = new Vector<xtandem_param>();
 		this.groups = new Vector<String>();
+		this.load_new_param();
 	}
 
 	public void load_xml(IdentificationXtandemFile xml) throws Exception {
@@ -30,7 +31,7 @@ public class xtandem_params {
 		new parse_file(new xml_xtandem_param(this), xml);
 	}
 
-	public void load_new_param() throws Exception {
+	private void load_new_param() throws Exception {
 		if (template == null)
 			throw new MSMSException(
 					"Problem accessing resource : xtandem_model.xml");

xtandempipeline/src/fr/inra/pappso/xtandempipeline/thread/combined.java

@@ -14,6 +14,7 @@ import fr.inra.pappso.xtandempipeline.filter_print.filter_recalcul_protein_evalu
 import fr.inra.pappso.xtandempipeline.filter_print.filter_unique_peptide_number;
 import fr.inra.pappso.xtandempipeline.filter_print.filter_unique_peptide_number_to;
 import fr.inra.pappso.xtandempipeline.filter_print.filter_peptide_evalue;
+import fr.inra.pappso.xtandempipeline.sax_parse.Mascot;
 import fr.inra.pappso.xtandempipeline.sax_parse.parse_file;
 import fr.inra.pappso.xtandempipeline.sax_parse.xtandem;
 
@@ -21,10 +22,6 @@ public class combined extends process {
 
 	private Vector<IdentificationDataFile> xtandem;
 
-	// private ArrayList<Identification> identifications =
-	// XtandemPipelineSession
-	// .getInstance().getIdentificationList();
-
 	public combined(config conf) {
 		super(conf);
 	}
@@ -34,12 +31,16 @@ public class combined extends process {
 
 		// Ecrire le process et penser à relever des StopException
 		xtandem handler = new xtandem();
+		Identification ident =null;
 		for (int i = 0; i < xtandem.size(); i++) {
 			// parsing
 			this.set_view("Parsing file : " + (i + 1) + "/" + xtandem.size()
 					+ "\n" + xtandem.get(i).getName());
 			handler.set_sample(xtandem.get(i));
 			new parse_file(handler, xtandem.get(i));
+			//Mascot mascot = new Mascot(ident, xtandem.get(i));
+			//mascot.readAndParseMascotFile();
+			//ident =mascot.getIdentification();
 
 			// verification pas arreté
 			if (this.isStop())
@@ -49,7 +50,7 @@ public class combined extends process {
 
 		// filtering
 		this.set_view("Parsing finished\n" + "Filtering peptides");
-		Identification ident = handler.get_identifications();
+		ident = handler.get_identifications();
 
 		// on mets à jour les liens
 		ident.filtered_match(new filter_index_and_order_match_scan());

xtandempipeline/src/fr/inra/pappso/xtandempipeline/thread/xtandem_analysis.java

@@ -46,7 +46,6 @@ public class xtandem_analysis extends process {
 		// Correction du fichier de parametres
 		logger.info("Corrected xtandem preset error");
 		xtandem_params params = new xtandem_params();
-		params.load_new_param();
 		params.load_xml(model);
 		params.save_to_xml(model);
 

xtandempipeline/src/fr/inra/pappso/xtandempipeline/ui/swt/Filter_Windows.java

@@ -26,7 +26,7 @@ import fr.inra.pappso.xtandempipeline.thread.DataType_Base;
 
 public class Filter_Windows {
 	protected config conf;
-	private Vector<IdentificationDataFile> add_xml;
+	private Vector<IdentificationDataFile> addDataFile;
 	private String current_directory;
 	private Shell shell;
 	private Text pep_evalue;
@@ -40,7 +40,7 @@ public class Filter_Windows {
 		Cancelled = true;
 		this.conf = XtandemPipelineSession.getInstance().getConfig();
 		this.type = XtandemPipelineSession.getInstance().getDataTypeBase();
-		this.add_xml = new Vector<IdentificationDataFile>();
+		this.addDataFile = new Vector<IdentificationDataFile>();
 		current_directory = conf.getXmlresult().getAbsolutePath();
 		shell = new Shell(controlleur.getDisplay(), SWT.DIALOG_TRIM
 				| SWT.APPLICATION_MODAL);
@@ -202,7 +202,7 @@ public class Filter_Windows {
 		try {
 			this.modified_conf();
 			conf.savePropertiestoXml();
-			type.getPro().setXtandem(add_xml);
+			type.getPro().setXtandem(addDataFile);
 			Cancelled = false;
 			shell.dispose();
 		} catch (MSMSException e) {
@@ -247,18 +247,18 @@ public class Filter_Windows {
 
 	private void add_clic(Button but) {
 		try {
-			IdentificationDataFile[] files = this.get_file_xml();
+			IdentificationDataFile[] files = this.getDataFile();
 			for (int m = 0; m < files.length; m++) {
-				if (!add_xml.contains(files[m]))
-					add_xml.add(files[m]);
+				if (!addDataFile.contains(files[m]))
+					addDataFile.add(files[m]);
 			}
 			// for (int i = 0; i < files.length; i++)
 			// add_xml.add(files[i]);
 			Text t = (Text) but.getData("text");
-			t.setText(add_xml.size() + " files selected");
+			t.setText(addDataFile.size() + " files selected");
 			t.update();
 
-			if (add_xml.size() > 0)
+			if (addDataFile.size() > 0)
 				((Button) but.getData("next")).setEnabled(true);
 
 		} catch (StopException S) {
@@ -290,10 +290,10 @@ public class Filter_Windows {
 		return (new File(fileName));
 	}
 
-	private IdentificationDataFile[] get_file_xml() throws Exception {
+	private IdentificationDataFile[] getDataFile() throws Exception {
 		FileDialog dlg = new FileDialog(shell, SWT.OPEN | SWT.MULTI);
-		dlg.setFilterExtensions(new String[] { "*.xml", "*.*" });
-		dlg.setFilterNames(new String[] { "X!Tandem result files", "All files" });
+		dlg.setFilterExtensions(new String[] { "*.xml", "*.dat","*.*" });
+		dlg.setFilterNames(new String[] { "X!Tandem result files", "Mascot result files","All files" });
 		dlg.setFilterPath(current_directory);
 		dlg.setText("Add new xml result files to analyse");
 		String fileName = dlg.open();
@@ -313,7 +313,7 @@ public class Filter_Windows {
 	}
 
 	public Vector<IdentificationDataFile> get_Add_xml() {
-		return add_xml;
+		return addDataFile;
 	}
 
 	public boolean isCancelled() {

xtandempipeline/src/fr/inra/pappso/xtandempipeline/ui/swt/Xtandem_preset.java

@@ -193,74 +193,78 @@ public class Xtandem_preset {
 	}
 
 	private void set_tab_folder() {
-		params = new xtandem_params();
 		try {
-			params.load_new_param();
-		} catch (Exception e) {
-			e.printStackTrace();
-		}
-		text_field = new Vector<Text>();
-		combo_field = new Vector<Combo>();
-
-		for (String groupname : params.get_groups()) {
-			TabItem item = new TabItem(tabFolder, SWT.NONE);
-			item.setText(groupname);
-			Composite grid = new Composite(tabFolder, SWT.NONE);
-			GridLayout layout = new GridLayout(3, false);
-			layout.verticalSpacing = 5;
-			layout.horizontalSpacing = 5;
-			grid.setLayout(layout);
-			for (xtandem_param param : params.get_params_to_group(groupname)) {
-				Label l = new Label(grid, SWT.SINGLE | SWT.RIGHT);
-				l.setText(param.toString());
-				l.setLayoutData(new GridData(SWT.FILL, SWT.CENTER, false, false));
-				if (param.get_list_item().length == 1 || !param.isModified()) {
-
-					Text text = new Text(grid, SWT.SINGLE);
-					text.setLayoutData(new GridData(GridData.FILL_HORIZONTAL));
-					text.setData("param", param);
-					if (!param.isModified()) {
-						text.setEditable(false);
-						text.setEnabled(false);
+			params = new xtandem_params();
+			text_field = new Vector<Text>();
+			combo_field = new Vector<Combo>();
+
+			for (String groupname : params.get_groups()) {
+				TabItem item = new TabItem(tabFolder, SWT.NONE);
+				item.setText(groupname);
+				Composite grid = new Composite(tabFolder, SWT.NONE);
+				GridLayout layout = new GridLayout(3, false);
+				layout.verticalSpacing = 5;
+				layout.horizontalSpacing = 5;
+				grid.setLayout(layout);
+				for (xtandem_param param : params
+						.get_params_to_group(groupname)) {
+					Label l = new Label(grid, SWT.SINGLE | SWT.RIGHT);
+					l.setText(param.toString());
+					l.setLayoutData(new GridData(SWT.FILL, SWT.CENTER, false,
+							false));
+					if (param.get_list_item().length == 1
+							|| !param.isModified()) {
+
+						Text text = new Text(grid, SWT.SINGLE);
+						text.setLayoutData(new GridData(
+								GridData.FILL_HORIZONTAL));
+						text.setData("param", param);
+						if (!param.isModified()) {
+							text.setEditable(false);
+							text.setEnabled(false);
+						}
+						text.setText(param.getValue());
+						text_field.add(text);
+					} else {
+						Combo com = new Combo(grid, SWT.DROP_DOWN
+								| SWT.READ_ONLY);
+						for (String value : param.get_list_item())
+							com.add(value);
+						com.select(param.getValueindex());
+						com.setData("param", param);
+						com.setLayoutData(new GridData(GridData.FILL_HORIZONTAL));
+						this.combo_field.add(com);
 					}
-					text.setText(param.getValue());
-					text_field.add(text);
-				} else {
-					Combo com = new Combo(grid, SWT.DROP_DOWN | SWT.READ_ONLY);
-					for (String value : param.get_list_item())
-						com.add(value);
-					com.select(param.getValueindex());
-					com.setData("param", param);
-					com.setLayoutData(new GridData(GridData.FILL_HORIZONTAL));
-					this.combo_field.add(com);
-				}
-				Button help = new Button(grid, SWT.PUSH);
-				help.setText("?");
-				help.setData("link", param.getLink());
-				help.addSelectionListener(new SelectionAdapter() {
-					public void widgetSelected(SelectionEvent e) {
-						InputStreamReader reader = new InputStreamReader(this
-								.getClass().getResourceAsStream(
-										(String) e.widget.getData("link")));
-						BufferedReader br = new BufferedReader(reader);
-						String str;
-						StringBuffer sb = new StringBuffer();
-						try {
-							while ((str = br.readLine()) != null) {
-								sb.append(str);
+					Button help = new Button(grid, SWT.PUSH);
+					help.setText("?");
+					help.setData("link", param.getLink());
+					help.addSelectionListener(new SelectionAdapter() {
+						public void widgetSelected(SelectionEvent e) {
+							InputStreamReader reader = new InputStreamReader(
+									this.getClass().getResourceAsStream(
+											(String) e.widget.getData("link")));
+							BufferedReader br = new BufferedReader(reader);
+							String str;
+							StringBuffer sb = new StringBuffer();
+							try {
+								while ((str = br.readLine()) != null) {
+									sb.append(str);
+								}
+								br.close();
+							} catch (IOException e1) {
+								e1.printStackTrace();
 							}
-							br.close();
-						} catch (IOException e1) {
-							e1.printStackTrace();
+							help_param.setText(sb.toString());
 						}
-						help_param.setText(sb.toString());
-					}
-				});
+					});
+				}
+				grid.pack();
+				item.setControl(grid);
 			}
-			grid.pack();
-			item.setControl(grid);
+			tabFolder.pack();
+		} catch (Exception e) {
+			this.view_error(e, "Problems during loading default params");
 		}
-		tabFolder.pack();
 	}
 
 	private void delete_xml_file() {

xtandempipeline/src/test/MascotTest.java

+package test;
+
+import java.util.Vector;
+
+import org.apache.log4j.Logger;
+
+import com.compomics.mascotdatfile.util.interfaces.Modification;
+import com.compomics.mascotdatfile.util.mascot.MascotDatfile;
+import com.compomics.mascotdatfile.util.mascot.PeptideHit;
+import com.compomics.mascotdatfile.util.mascot.Query;
+import com.compomics.mascotdatfile.util.mascot.QueryToPeptideMap;
+
+import fr.inra.pappso.xtandempipeline.class_msms.IdentificationDataFile.IdentificationDataFile;
+import fr.inra.pappso.xtandempipeline.sax_parse.Mascot;
+
+public class MascotTest {
+	// Class specific log4j logger for TestMascotDatfile instances.
+	private static Logger logger = Logger.getLogger(MascotTest.class);
+
+	public static void main(String[] args) {
+		String ds = "/tmp/FtoeoxSSO.dat";
+		//String ds = "/media/E860-74C3/Triplex1_FT_light.dat";
+		try {
+			IdentificationDataFile data = IdentificationDataFile.newIdentificationDataFile(ds);
+			Mascot mascot = new Mascot(null,data);
+			mascot.readAndParseMascotFile();
+//			MascotDatfile mascot = new MascotDatfile(ds);
+//			logger.info("Read :" + mascot.getFileName());
+//			QueryToPeptideMap map = mascot.getQueryToPeptideMap();
+//
+//			for (Object q : mascot.getQueryList()) {
+//				Query query = (Query) q;
+//				// System.out.println(query.getScans() + " | "
+//				// + query.getRetentionTimeInSeconds() + " | "
+//				// + query.getQueryNumber());
+//				Vector<?> peptides = map.getAllPeptideHits(query
+//						.getQueryNumber());
+//				if (peptides != null) {
+//					for (Object ob : peptides) {
+//						PeptideHit pep = (PeptideHit) ob;
+//						if (pep.getExpectancy() < 0.05) {
+//							System.out.println(query.getScans() + " | "
+//									+ query.getRetentionTimeInSeconds() + " | "
+//									+ query.getQueryNumber()+ " | "+query.getChargeString());
+//							System.out.println(query.getChargeString()+" | "+(int) round((pep.getPeptideMr()/query.getPrecursorMZ()),0));
+//							
+//							System.out.println(pep.getModifiedSequence() + " | "
+//									+ pep.getIonsScore() + " | "
+//									+ pep.getHomologyThreshold());
+//							for (Modification mod : pep.getModifications()) {
+//								if (mod != null)
+//									System.out.println(mod.getLocation() + " | "
+//											+ mod.getMass() + " | " + mod.getShortType());
+//							}
+////							System.out.println();
+////							for(String i : pep.getModifiedSequenceComponents()){
+////								System.out.println(i);
+////							}
+//						}
+//					}
+//				}
+			//}
+
+		} catch (Exception e) {
+			// TODO Auto-generated catch block
+			e.printStackTrace();
+		}
+	}
+	public static double round(double what, int howmuch) {
+		return (double) ((int) (what * Math.pow(10, howmuch) + .5))
+				/ Math.pow(10, howmuch);
+	}
+}