Changeset 4535

Timestamp:

Nov 22, 2013, 4:39:31 PM (9 years ago)

Author:

Fredrik Levander

Message:

Refs #819. Adding support for matching low-confidence identifications to features in presence of high-scoring hits with same sequence. Setting clusterIds for features with sequence (same cluster id for same sequence and charge).

Location:

trunk

Files:

• api/core/conf/common-queries.xml (modified) (1 diff)
• api/core/src/org/proteios/core/Hit.java (modified) (1 diff)
• client/servlet/src/locale/en/dictionary (modified) (1 diff)
• client/servlet/src/org/proteios/action/feature/CreateFeatureHitMatchJob.java (modified) (3 diffs)
• client/servlet/src/org/proteios/action/feature/FeatureHitMatchForm.java (modified) (1 diff)
• plugin/src/org/proteios/plugins/FeatureHitMatcher.java (modified) (17 diffs)

trunk/api/core/conf/common-queries.xml

@@ -255 +255 @@
     </description>
   </query>
+
+  <query id="GET_UNIQUE_PEPTIDE_SEQUENCES_BELOW_FDR_IN_HITS_FOR_PROJECT" type="HQL">
+    <sql>
+      SELECT DISTINCT h.description
+      FROM HitData h
+      WHERE h.project = :project AND h.protein = false AND h.combinedFDR &lt;= :combinedFDR
+      ORDER by h.description 
+    </sql>
+    <description>
+      Load all distinct (unique) peptide sequences in the Hits table for a project.
+    </description>
+  </query>
   
   <query id="GET_UNIQUE_CHARGES_IN_HITS_FOR_PROJECT" type="HQL">

trunk/api/core/src/org/proteios/core/Hit.java

@@ -145 +145 @@
       "GET_UNIQUE_PEPTIDE_SEQUENCES_IN_HITS_FOR_PROJECT");
     query.setEntity("project", project.getData());
+    return query.list();
+  }
+  
+  @SuppressWarnings("unchecked")
+  public static List<String> getUniquePeptideSequencesBelowFDR(Project project,
+      DbControl dc,float combinedFDR)
+  {
+    org.hibernate.Query query = HibernateUtil.getPredefinedQuery(dc
+      .getHibernateSession(),
+      "GET_UNIQUE_PEPTIDE_SEQUENCES_BELOW_FDR_IN_HITS_FOR_PROJECT");
+    query.setEntity("project", project.getData());
+    query.setFloat("combinedFDR", combinedFDR);
     return query.list();
   }

trunk/client/servlet/src/locale/en/dictionary

@@ -848 +848 @@
 SearchXTandemUsingLocalInstallation=Search X!Tandem using local installation
 SearchXTandemViaWebInterface=Search X!Tandem via web interface
+SecondaryPeptideCutOff=Secondary peptide cutoff
 SelectFileType=Select File Type
 SelectFractionIdEntryMode=Fraction ID input:

trunk/client/servlet/src/org/proteios/action/feature/CreateFeatureHitMatchJob.java

@@ -57 +57 @@
   public static final VFloat VALIGNTOL = new VFloat("aligntol",true);
   public static final VFloat VFDRCUTOFF = new VFloat("fdrCutOff",true);
+  public static final VFloat VSECONDARYFDRCUTOFF = new VFloat("secondaryFdrCutOff",true);
   
   /*
@@ -78 +79 @@
     Float alignTimeTol = getValidFloat(VALIGNTOL);
     Float fdrCutOff = getValidFloat(VFDRCUTOFF);
+    Float secondaryFdrCutOff = getValidFloat(VSECONDARYFDRCUTOFF);
     Boolean isSecondary = getValidBoolean(VSECONDARY);
     if (isSecondary == null)
@@ -115 +117 @@
       isSecondary);
     job.setParameterValue("fdrCutOff", new FloatParameterType(), fdrCutOff);
+    job.setParameterValue("secondaryFdrCutOff", new FloatParameterType(), secondaryFdrCutOff);
     job.setName("Match features and hits");
     job.setDescription("Matching in project:" + project+ ", FDR cutoff:" + fdrCutOff + ", tolerance:"+tolerance + ", secondary:"+isSecondary);

trunk/client/servlet/src/org/proteios/action/feature/FeatureHitMatchForm.java

@@ -57 +57 @@
         CreateFeatureHitMatchJob.VFDRCUTOFF);
     tolFDRF.setValue(new Float(0.01));
-    tolFDRF.setLabel("peptideCutOff");
+    tolFDRF.setLabel("PeptideCutOff");
     tolFDRF.setHelp("Peptide identification FDR cutoff.");
     properties.add(tolFDRF);
+    TextField<Float> stolFDRF = new TextField<Float>(
+        CreateFeatureHitMatchJob.VSECONDARYFDRCUTOFF);
+    stolFDRF.setValue(new Float(1f));
+    stolFDRF.setLabel("SecondaryPeptideCutOff");
+    stolFDRF.setHelp("FDR cutoff for peptides that pass the primary FDR cutoff elsewhere in the project.");
+    properties.add(stolFDRF);
     Checkbox<VBoolean> isSecondaryF = new Checkbox<VBoolean>(
         CreateFeatureHitMatchJob.VSECONDARY);

trunk/plugin/src/org/proteios/plugins/FeatureHitMatcher.java

@@ -61 +61 @@
 import java.io.OutputStream;
 import java.io.PrintWriter;
+import java.sql.Timestamp;
 import java.util.ArrayList;
 import java.util.Arrays;
+import java.util.Date;
 import java.util.List;
+import java.util.TreeMap;
 
 /**
@@ -76 +79 @@
   static float MONOISOTOPIC_PROTON_MASS = (float) 1.007276035;
   boolean jobAborted = false;
+  long currentCluster =0;
+  TreeMap<String,Long> clusterMap = new TreeMap<String,Long>();
 
   /**
@@ -93 +98 @@
   public About getAbout() {
     return new AboutImpl("Match features to hits",
-        "Match features to hits for a project", "0.4",
-        "2009-2012, Marianne Sandin, Fredrik Levander", null, null,
+        "Match features to hits for a project", "0.5",
+        "2009-2013, Marianne Sandin, Fredrik Levander", null, null,
         "http://www.proteios.org");
   }
@@ -113 +118 @@
       double alignTimeTol = 4;
       float fdrCutOff = 0.01f;
+      float secondaryFdrCutOff = 1f;
       int count = 0;
       DbControl dc = sc.newDbControl();   
@@ -146 +152 @@
       alignTimeTol = ((Float) job.getValue("alignTimeTol")).doubleValue();
       fdrCutOff = ((Float) job.getValue("fdrCutOff")).floatValue();
+      secondaryFdrCutOff = ((Float) job.getValue("secondaryFdrCutOff")).floatValue();
+
       boolean matchRelated = ((Boolean) job.getValue("secondary"))
           .booleanValue();
@@ -165 +173 @@
 
         writer.println("Hit FDR cutoff:" + fdrCutOff);
+        writer.println("Secondary hit FDR cutoff:" + secondaryFdrCutOff);
         writer.println("m/z tolerance:" + tolerance);
         writer.println("RT (minutes) tolerance - start:" + startTimeTol
             + " end:" + endTimeTol);
-        writer.println("Matching with related files" + matchRelated);
+        writer.println("Matching with related files:" + matchRelated);
         
         if (matchRelated) {
@@ -185 +194 @@
             + "\n");
         ItemQuery<Feature> featureQuery = qf.select(Feature.class);
+        List<String> peptideSequenceList = Hit.getUniquePeptideSequencesBelowFDR(project, dc, fdrCutOff);
+        for (int i=0;i<peptideSequenceList.size();i++)
+        {
+          String s = peptideSequenceList.get(i);
+          if (s.contains("delta")) {
+            peptideSequenceList.set(i,s.substring(0, s.indexOf("delta")));
+          }
+        }
+        Timestamp tStmp = new Timestamp(new Date().getTime());
+        currentCluster = tStmp.getTime();
         featureQuery.restrictPermanent(Restrictions.eq(
             Hql.property("project"), Hql.entity(project)));
@@ -215 +234 @@
           writer.println("File:" + currentMsFile);
           List<Hit> hits = retrieveHits(currentMsFile, project, dc2,
-              fdrCutOff, matchRelated);
+              fdrCutOff, secondaryFdrCutOff, peptideSequenceList, matchRelated);
           List<Feature> features = featureQuery.list(dc2);
           PolynomialSplineFunction func = null;
@@ -387 +406 @@
           log.debug("m/z matching:" + f.getMassToChargeRatio()
               + " hit:" + hitMz);
-          // In rare cases the end retention is null!
+          // In rare cases the start or end retention time can be null!
           if (f.getEndRetentionTimeInMinutes() == null)
             f.setEndRetentionTimeInMinutes(new Float(f
                 .getApexRetentionTimeInMinutes() + endTimeTol));
+          if (f.getStartRetentionTimeInMinutes() == null)
+            f.setStartRetentionTimeInMinutes(new Float(f
+              .getApexRetentionTimeInMinutes() - startTimeTol));
           if (hitRt >= f.getStartRetentionTimeInMinutes()
               && hitRt <= f.getEndRetentionTimeInMinutes()) {
@@ -413 +435 @@
                 replaced++;
               }
-              f.setPeptideSequence(h.getDescription().substring(0,h.getDescription().contains("delta") ? h.getDescription().lastIndexOf(" ") : h.getDescription().length()));
+              setFeatureSequenceAndClusterId(f, h);
               f.addHit(h);
               newmatches++;
@@ -470 +492 @@
                   if (best) {
                     
-                    f.setPeptideSequence(h.getDescription().substring(0, h.getDescription().contains("delta") ? h.getDescription().lastIndexOf(" ") : h.getDescription().length()));
+                    setFeatureSequenceAndClusterId(f, h);
                     writer.println("Better scoring hit sequence replacing old");
                     // We only want the precursor quantity
@@ -526 +548 @@
               h.setPrecursorQuantity(f.getIntegratedIntensity());
               f.addHit(h);
-              f.setPeptideSequence(h.getDescription().substring(0, h.getDescription().contains("delta") ? h.getDescription().lastIndexOf(" ") : h.getDescription().length()));
+              setFeatureSequenceAndClusterId(f, h);
               count++;
               newmatches++;
@@ -579 +601 @@
     return count;
   }
+  
+  private void setFeatureSequenceAndClusterId(Feature f, Hit h)
+  {
+    f.setPeptideSequence(h.getDescription().substring(0,h.getDescription().contains("delta") ? h.getDescription().lastIndexOf(" ") : h.getDescription().length()));
+    String comb = f.getPeptideSequence()+","+f.getChargeState();
+    if (clusterMap.containsKey(comb))
+    {
+      f.setClusterId(clusterMap.get(comb));
+    }
+    else
+    {
+      f.setClusterId(currentCluster);
+      clusterMap.put(comb, currentCluster);
+      currentCluster++;
+    }
+  }
 
   private int getFirstHitForChargeState(List<Hit> hits, int charge) {
@@ -590 +628 @@
 
   private List<Hit> retrieveHits(File msFile, Project project, DbControl dc,
-      float fdrCutOff, boolean matchRelated) {
+      float fdrCutOff, float secondaryFdrCutOff, List<String> peptideSequenceList, boolean matchRelated) {
     List<Hit> hits = new ArrayList<Hit>();
     ItemQuery<Hit> hitQuery = Hit.getQuery(project);
+    int notSoGoodHits = 0;
     // First check if the same msXML file can be found in the hits table
     hitQuery.restrict(Restrictions.eq(Hql.property("peakListFile"),
@@ -601 +640 @@
       hitQuery.setPermanentParameter("protein", false, null);
       hitQuery.restrictPermanent(Restrictions.lteq(
-          Hql.property("combinedFDR"), Expressions.aFloat(fdrCutOff)));
+          Hql.property("combinedFDR"), Expressions.aFloat(secondaryFdrCutOff)));
     }
     if (hitQuery.count(dc) == 0) {
@@ -668 +707 @@
         hitQuery.restrictPermanent(Restrictions.lteq(
             Hql.property("combinedFDR"),
-            Expressions.aFloat(fdrCutOff)));
+            Expressions.aFloat(secondaryFdrCutOff)));
         log.debug("Hit peakListFile:" + h.getPeakListFile());
       }
@@ -676 +715 @@
     ItemResultList<Hit> hitlist = hitQuery.list(dc);
     for (Hit h : hitlist) {
-      hits.add(h);
-    }
-    log.debug("Hits found:" + hits.size());
+      if (h.getCombinedFDR()<=fdrCutOff)
+        hits.add(h);
+      else {
+        String s = h.getDescription();
+        if (s.contains("delta")) 
+        {
+          s = s.substring(0, s.indexOf("delta"));
+        }
+        if (peptideSequenceList.contains(s))
+        {
+          hits.add(h);
+          notSoGoodHits++;
+        }
+      }
+    }
+    log.debug("Hits found:" + hits.size() + ". " + notSoGoodHits + " retained above primary FDR threshold.");
     return hits;
   }