PAI computing method checked

All PAI computed in xtp are using the same method : number of spectrums from valid peptides divided by the theoretical number of tryptic peptides

src/fr/inra/pappso/xtandempipeline/class_msms/Match.java

@@ -120,17 +120,69 @@ public class Match implements Comparable<Match>, GrpProtMatch {
 		return (samples);
 	}
 
-	public int get_spectra_numbers_to(MsRun samp) throws MSMSException {
+	/**
+	 * count number of spectra coming from one MsRun (MS sample)
+	 * 
+	 * @param samp
+	 *            the sample to count spectra from
+	 * @return
+	 * @throws MSMSException
+	 */
+	private int getSpectraNumberIn(MsRun samp) throws MSMSException {
 		int i = 0;
 		for (Peptide pep : peptides_match.values()) {
-			if (pep.getMsRun().equals(samp))
+			if (pep.isValidate()) {
+				if (pep.getMsRun().equals(samp)) {
+					i++;
+				}
+			}
+		}
+		return (i);
+	}
+
+	/**
+	 * count number of spectra in this match
+	 * 
+	 * @return integer number of spectra
+	 * @throws MSMSException
+	 */
+	private int getSpectraNumber() {
+		int i = 0;
+		for (Peptide pep : peptides_match.values()) {
+			if (pep.isValidate()) {
 				i++;
+			}
 		}
+
 		return (i);
 	}
 
-	public int get_spectra_numbers() {
-		return (this.peptides_match.size());
+	/**
+	 * compute the Protein Abundance Index (PAI)
+	 * 
+	 * @return
+	 */
+	public double getPAI() {
+		// compute PAI
+		double value = (double) this.getSpectraNumber()
+				/ (double) this.get_protein_match()
+						.getNumberOfTrypticPeptidesForPAI();
+
+		return value;
+	}
+
+	/**
+	 * compute the Protein Abundance Index (PAI) within an MsRun
+	 * 
+	 * @return
+	 */
+	public double getPAIbyMsRun(MsRun samp) throws MSMSException {
+		// compute PAI
+		double value = (double) this.getSpectraNumberIn(samp)
+				/ (double) this.get_protein_match()
+						.getNumberOfTrypticPeptidesForPAI();
+
+		return value;
 	}
 
 	public int get_unique_valid_peptide_number_to(MsRun samp)
@@ -161,8 +213,8 @@ public class Match implements Comparable<Match>, GrpProtMatch {
 	}
 
 	public ArrayList<Peptide> get_peptide_order() {
-		ArrayList<Peptide> peptides = new ArrayList<Peptide>(peptides_match
-				.values());
+		ArrayList<Peptide> peptides = new ArrayList<Peptide>(
+				peptides_match.values());
 		Collections.sort(peptides);
 		return (peptides);
 	}

src/fr/inra/pappso/xtandempipeline/class_msms/Protein.java

@@ -33,7 +33,7 @@ public class Protein implements GrpProtein {
 	private String dbxref_type;
 
 	private String dbxref_accession;
-	
+
 	private boolean isReversed;
 
 	// constructeur
@@ -70,10 +70,10 @@ public class Protein implements GrpProtein {
 			logger.debug("Corrected reverse protein description");
 			this.description = S.replaceFirst("\\:reversed$", "");
 			this.setReversed(true);
-		} else if(S.split(" ")[0].matches(".*\\|reversed$")){
-			this.isReversed=true;
+		} else if (S.split(" ")[0].matches(".*\\|reversed$")) {
+			this.isReversed = true;
 			this.description = S;
-		}else
+		} else
 			this.description = S;
 		this.search_dbxref();
 		// logger.debug("set_description end");
@@ -115,10 +115,6 @@ public class Protein implements GrpProtein {
 		return (mw);
 	}
 
-	public int get_PAI_count() {
-		return (PAI_count);
-	}
-
 	public String get_description() {
 		return (description);
 	}
@@ -215,6 +211,12 @@ public class Protein implements GrpProtein {
 
 	}
 
+	/**
+	 * get the theoretical number of tryptic peptides between 800 and 2500
+	 * Daltons for one protein
+	 * 
+	 * @return integer number of possible tryptic peptides
+	 */
 	private int determined_PAI_count() {
 		int count = 0;
 		Hashtable<String, String> aa = new Hashtable<String, String>();
@@ -326,13 +328,18 @@ public class Protein implements GrpProtein {
 	public boolean isReversed() {
 		return isReversed;
 	}
+
 	public void setReversed(boolean rev) {
-		this.isReversed=rev;
-		//forced reverse annotation on protein accession
-		if(isReversed){
+		this.isReversed = rev;
+		// forced reverse annotation on protein accession
+		if (isReversed) {
 			this.description = description.replaceFirst(" ", "|reversed ");
 		}
 	}
-	
+
+	@Override
+	public int getNumberOfTrypticPeptidesForPAI() {
+		return this.PAI_count;
+	}
 
 }

src/fr/inra/pappso/xtandempipeline/filter_print/swt_table_protein.java

@@ -80,8 +80,8 @@ public class swt_table_protein extends swt_table_base {
 				if (filter.equals(""))
 					is_filter = true;
 				else {
-					test = pattern.matcher(((Match)hashProt.getGrpProtMatch())
-								.get_protein_match().get_description());
+					test = pattern.matcher(((Match) hashProt.getGrpProtMatch())
+							.get_protein_match().get_description());
 					if (test.matches())
 						is_filter = true;
 					// if (hashProt.getMatch().get_protein_match()
@@ -122,7 +122,8 @@ public class swt_table_protein extends swt_table_base {
 						int contain_multi = 0;
 						for (HashSampleScan hash : phospho) {
 							if (phospho.getGrpProtMatch()
-									.getBestPeptideToHashSampleScan(hash).getGrpPhosphoModifs().size() > 1) {
+									.getBestPeptideToHashSampleScan(hash)
+									.getGrpPhosphoModifs().size() > 1) {
 								contain_multi++;
 							}
 						}
@@ -145,8 +146,7 @@ public class swt_table_protein extends swt_table_base {
 						}
 						item.setText(6,
 								"" + hashProt.getUniquePeptideHashSampleScan());
-						item.setText(7, "" + (float) hashProt.size()
-								/ top.get_protein_match().get_PAI_count());
+						item.setText(7, "" + (float) hashProt.getPAI());
 					}
 					if (!color)
 						item.setBackground(table.getDisplay().getSystemColor(

src/fr/inra/pappso/xtandempipeline/grouping/HashSampleScanSetProt.java

@@ -148,4 +148,34 @@ public class HashSampleScanSetProt extends HashSampleScanSet {
 		}
 		return (grpMsSampleList);
 	}
+
+	/**
+	 * compute the Protein Abundance Index (PAI)
+	 * 
+	 * @return
+	 */
+	public double getPAI() {
+		// compute PAI
+		double value = (double) this.getTotalHashSampleScan()
+				/ (double) this.match.getGrpProtein()
+						.getNumberOfTrypticPeptidesForPAI();
+
+		return value;
+	}
+
+	/**
+	 * compute the Protein Abundance Index (PAI) within an MsRun
+	 * 
+	 * @return
+	 */
+	public double getPAIbyMsRun(GrpMsSample samp) throws MSMSException {
+		// compute PAI
+
+		double value = (double) getTotalHashSampleScantoMsRun(samp)
+				/ (double) this.match.getGrpProtein()
+						.getNumberOfTrypticPeptidesForPAI();
+
+		return value;
+	}
+
 }

src/fr/inra/pappso/xtandempipeline/grouping/interfaces/GrpProtein.java

@@ -3,8 +3,10 @@ package fr.inra.pappso.xtandempipeline.grouping.interfaces;
 public interface GrpProtein extends Comparable<GrpProtein> {
 
 	String getAccession();
-	
+
 	@Override
 	public int compareTo(GrpProtein arg0);
 
+	public int getNumberOfTrypticPeptidesForPAI();
+
 }

src/fr/inra/pappso/xtandempipeline/output/ProticDbMlWriter.java

@@ -467,11 +467,12 @@ public class ProticDbMlWriter {
 					+ XtandemPipelineSession.getInstance().getConfig()
 							.get_peptide_evalue(),
 					"X!TandemPipeline filter on peptide evalue");
-			
-			//TODO write database filter
-			//this.writeCvParam("PROTICdbO:0000324", ""
-			//		+ XtandemPipelineSession.getInstance().getConfig().get_database_filter(),
-			//		"X!TandemPipeline filter on peptide evalue");
+
+			// TODO write database filter
+			// this.writeCvParam("PROTICdbO:0000324", ""
+			// +
+			// XtandemPipelineSession.getInstance().getConfig().get_database_filter(),
+			// "X!TandemPipeline filter on peptide evalue");
 
 			writer.writeEndElement();// processingMethod
 
@@ -766,7 +767,7 @@ public class ProticDbMlWriter {
 						// id: PROTICdbO:0000335
 						// name: X!TandemPipeline PAI
 						this.writeCvParam("PROTICdbO:0000335",
-								"" + prot.get_PAI_count(), "PAI");
+								"" + hashProt.getPAI(), "PAI");
 
 						// [Term]
 						// id: PROTICdbO:0000337

src/fr/inra/pappso/xtandempipeline/output/SpreadSheets/OdsComparPaiOutput.java

@@ -190,10 +190,8 @@ public class OdsComparPaiOutput {
 
 			for (String msSampleName : msRunOrderedList.keySet()) {
 				// compute PAI
-				double value = (double) topMatch
-						.get_spectra_numbers_to(this.msRunOrderedList
-								.get(msSampleName))
-						/ (double) topMatch.get_protein_match().get_PAI_count();
+				double value = sg.getHashSampleScanSetProtList().get(0)
+						.getPAIbyMsRun(this.msRunOrderedList.get(msSampleName));
 
 				if (value > 0) {
 					odsTable.writeCell(value);
@@ -203,6 +201,15 @@ public class OdsComparPaiOutput {
 
 					// S.append("\t");
 				}
+				/*
+				 * value = topMatch.getPAIbyMsRun(this.msRunOrderedList
+				 * .get(msSampleName));
+				 * 
+				 * if (value > 0) { odsTable.writeCell(value); // S.append("\t"
+				 * + value); } else { odsTable.writeCell("");
+				 * 
+				 * // S.append("\t"); }
+				 */
 			}
 
 			logger.debug("writeOneSubGroup end");

src/fr/inra/pappso/xtandempipeline/output/SpreadSheets/OdsProteinOutput.java

@@ -172,13 +172,14 @@ public class OdsProteinOutput {
 			logger.debug("writeOneProtein begin " + group + sg);
 			odsTable.writeLine();
 
-			Protein prot = (Protein) hashProt.getGrpProtMatch().getGrpProtein();
+			Match match = (Match) hashProt.getGrpProtMatch();
+
+			Protein prot = match.getGrpProtein();
 			odsTable.writeCell(Utils.getPappsoGroupId(group));
 			odsTable.writeCell(Utils.getPappsoSubGroupId(group, sg));
 			odsTable.writeCell(Utils.getPappsoGroupingNumber(group, sg, k));
 			odsTable.writeCell(prot.get_description());
 			odsTable.writeCell(prot.get_evalue());
-			Match match = (Match) hashProt.getGrpProtMatch();
 			// odsTable.writeCell("Coverage");
 			odsTable.writeCell(match.get_coverage());
 			// odsTable.writeCell("MW");
@@ -199,8 +200,9 @@ public class OdsProteinOutput {
 			// odsTable.writeCell("Uniques");
 			odsTable.writeCell(hashProt.getUniquePeptideHashSampleScan());
 			// odsTable.writeCell("PAI");
-			odsTable.writeCell(new Double((float) hashProt
-					.getTotalHashSampleScan() / prot.get_PAI_count()));
+			//odsTable.writeCell(new Double((float) hashProt
+			//		.getTotalHashSampleScan() / prot.get_PAI_count()));
+			odsTable.writeCell(hashProt.getPAI());
 			/*
 			 * ancien PAI odsTable.writeCell(new Double((float)
 			 * top.get_spectra_numbers_to(sample) /