Changeset 4446

Timestamp:

Mar 13, 2013, 3:45:40 PM (10 years ago)

Author:

marianne

Message:

Refs #774. Tolerances for extracting matching features are calculated from the set of matching sequences.

File:

• trunk/plugin/src/org/proteios/plugins/FeatureSequencePropagator.java (modified) (7 diffs)

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

@@ -36 +36 @@
 import java.util.Comparator;
 import java.util.Date;
-import java.util.Iterator;
 import java.util.List;
 import java.util.ListIterator;
@@ -47 +46 @@
 import org.apache.commons.collections.ListUtils;
 import org.apache.commons.collections.functors.NotNullPredicate;
-import org.apache.commons.lang.ArrayUtils;
 import org.apache.commons.math.ArgumentOutsideDomainException;
 import org.apache.commons.math.MathException;
@@ -545 +543 @@
             
             if (totNbrMatches < MIN_NBR_MATCHING_SEQ) {
-
-              double mzForPointMatchByTol = 0.01;
-              double rtForPointMatchByTol = 5;
-
-              //TODO:fill up with tolerances matches, check for duplicates when finding tolerance matches instead of clearing the points list
-              points.clear();
-
+              
               writer.println("There are too few common sequences for file "
                   + (fileNbr + 1)
                   + " and "
-                  + (secFileNbr + 1) + ".");
-              writer.println("Alignment will be performed by using tolerances: " +mzForPointMatchByTol +" Da and " +rtForPointMatchByTol +" min.");
+                  + (secFileNbr + 1) + ". Adding landmarks by tolerances.");
               log.debug("There are too few common sequences for file "
                   + (fileNbr + 1)
                   + " and "
-                  + (secFileNbr + 1) + ".");
-              log.debug("Alignment will be performed by using tolerances: " +mzForPointMatchByTol +" Da and " +rtForPointMatchByTol +" min.");
+                  + (secFileNbr + 1) + ". Adding landmarks by tolerances.");
+              
+              
+              double mzForPointMatchByTol = 0.01;
+              double rtForPointMatchByTol = 10;
+              
+              
+              if (totNbrMatches >= 20){
+                
+                writer.println("Estimating tolerances from sequences.");
+                log.debug("Estimating tolerances from sequences.");
+                
+                int quartileFactor = 10;
+                double[] initialMzRTVec = getInitialMzAndRT(points, quartileFactor, writer);
+                
+                mzForPointMatchByTol = initialMzRTVec[0];
+                rtForPointMatchByTol = initialMzRTVec[1];
+                
+              }
+
+              points.clear();
+
+              
+              writer.println("Feature pairs will be extracted by using tolerances: " +mzForPointMatchByTol +" Da and " +rtForPointMatchByTol +" min.");
+              
+              log.debug("Feature pairs will be extracted by using tolerances: " +mzForPointMatchByTol +" Da and " +rtForPointMatchByTol +" min.");
 
               simValue = getPointMatchByTol(points,
@@ -573 +588 @@
               simTemp = simValue;
 
-              //TODO: integrate simValue for using both sequences and tolerances
               simValue = getPointMatchByTol(points,
                   uniqueMsFiles.get
@@ -596 +610 @@
               if (totNbrMatches < MIN_NBR_MATCHING_SEQ) {
 
-                writer.println("There are still too few common sequences for file "
+                writer.println("There are still too few common features for file "
                     + (fileNbr + 1)
                     + " and "
                     + (secFileNbr + 1) + ".");
                 writer.println("No alignment will be performed for this file-pair.");
-                log.debug("There are still too few common sequences for file "
+                log.debug("There are still too few common features for file "
                     + (fileNbr + 1)
                     + " and "
@@ -614 +628 @@
             // considered an outlier
             int quartileFactor = 10;
-            int nbrOfOutliers = 0;
-
-            Collections.sort(points, new DiffComparator());
-
-            double lowerQuartile = points.get(
-                (int) Math.round(0.25 * points.size()))
-                .getDiff();
-            double upperQuartile = points.get(
-                (int) Math.round(0.75 * points.size()))
-                .getDiff();
-            double medianQuartile = points.get(
-                (int) Math.round(0.5 * points.size()))
-                .getDiff();
-            double interQuartileRange = upperQuartile
-                - lowerQuartile;
-            double upperFence = upperQuartile + quartileFactor
-                * interQuartileRange;
-
-            writer.println("Median RT difference before alignment: "
-                + medianQuartile);
-            writer.println("Largest RT difference before alignment: "
-                + points.get(points.size() - 1).getDiff());
-
-            int maxCutOff = points.size() - 1;
-            while (points.get(maxCutOff).getDiff() > upperFence) {
-              points.remove(maxCutOff);
-              maxCutOff--;
-              nbrOfOutliers++;
-            }
-
-            writer.println("Removing "
-                + nbrOfOutliers
-                + " outliers. Total number of sequence matches is now: "
-                + points.size());
-
-            medianQuartile = points.get(
-                (int) Math.round(0.5 * points.size()))
-                .getDiff();
-
-            writer.println("Median RT difference is now: "
-                + medianQuartile);
-            writer.println("Largest RT difference is now: "
-                + points.get(points.size() - 1).getDiff());
-
-            Collections.sort(points, new MzDiffComparator());
-            double medianQuartileMz = points.get(
-                (int) Math.round(0.5 * points.size()))
-                .getMzDiff();
-            writer.println("Median mz difference before alignment: "
-                + medianQuartileMz);
-            writer.println("Largest mz difference before alignment: "
-                + points.get(points.size() - 1).getMzDiff());
-
-            // the mz tolrance is set as the largest (rounded) mz
-            double mzTol = points.get(points.size() - 1)
-                .getMzDiff();
-            mzTol = Math.round(100000 * mzTol + 1)
-                / (double) 100000;
-
-            log.debug("Mz tol is: " + mzTol);
-
-            Collections.sort(points, new RTComparator());
-
+            
+            double[] initialMzRTVec = getInitialMzAndRT(points, quartileFactor, writer);
+            double mzTol = initialMzRTVec[0];
+            
+            
             if (simValue == 0) {
               simValue = points.size() * 2.0
@@ -1471 +1427 @@
     }
 
+  }
+  
+  private double[] getInitialMzAndRT(ArrayList<Point> points, int quartileFactor, PrintWriter writer){
+    
+    double[] initialMzRTVec = new double[2];
+    
+    int nbrOfOutliers = 0;
+
+    Collections.sort(points, new DiffComparator());
+
+    double lowerQuartile = points.get(
+        (int) Math.round(0.25 * points.size()))
+        .getDiff();
+    double upperQuartile = points.get(
+        (int) Math.round(0.75 * points.size()))
+        .getDiff();
+    double medianQuartile = points.get(
+        (int) Math.round(0.5 * points.size()))
+        .getDiff();
+    double interQuartileRange = upperQuartile
+        - lowerQuartile;
+    double upperFence = upperQuartile + quartileFactor
+        * interQuartileRange;
+
+    writer.println("Median RT difference before alignment: "
+        + medianQuartile);
+    writer.println("Largest RT difference before alignment: "
+        + points.get(points.size() - 1).getDiff());
+
+    int maxCutOff = points.size() - 1;
+    while (points.get(maxCutOff).getDiff() > upperFence) {
+      points.remove(maxCutOff);
+      maxCutOff--;
+      nbrOfOutliers++;
+    }
+
+    writer.println("Removing "
+        + nbrOfOutliers
+        + " outliers. Total number of sequence matches is now: "
+        + points.size());
+
+    medianQuartile = points.get(
+        (int) Math.round(0.5 * points.size()))
+        .getDiff();
+
+    writer.println("Median RT difference is now: "
+        + medianQuartile);
+    
+    double largestRTDiff = points.get(points.size() - 1).getDiff();
+    
+    
+    writer.println("Largest RT difference is now: "
+        +largestRTDiff );
+    
+    largestRTDiff = Math.round(100000 * largestRTDiff + 1)
+        / (double) 100000;
+    
+    Collections.sort(points, new MzDiffComparator());
+    double medianQuartileMz = points.get(
+        (int) Math.round(0.5 * points.size()))
+        .getMzDiff();
+    writer.println("Median mz difference before alignment: "
+        + medianQuartileMz);
+    writer.println("Largest mz difference before alignment: "
+        + points.get(points.size() - 1).getMzDiff());
+
+    // the mz tolrance is set as the largest (rounded) mz
+    double mzTol = points.get(points.size() - 1)
+        .getMzDiff();
+    mzTol = Math.round(100000 * mzTol + 1)
+        / (double) 100000;
+
+    log.debug("Mz tol is: " + mzTol);
+    
+    initialMzRTVec[0]=mzTol;
+    initialMzRTVec[1]=largestRTDiff;
+
+    Collections.sort(points, new RTComparator());
+    
+    return initialMzRTVec;
   }