Changeset 4441

Timestamp:

Mar 12, 2013, 12:52:29 PM (10 years ago)

Author:

marianne

Message:

Refs #774. Added control to make sure that there are no unaligned subgroups of files before ending the alignment loop.

File:

• trunk/plugin/src/org/proteios/plugins/FeatureSequencePropagator.java (modified) (11 diffs)

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

@@ -40 +40 @@
 import java.util.ListIterator;
 import java.util.Map.Entry;
+import java.util.Set;
 import java.util.TreeMap;
 import java.util.TreeSet;
@@ -167 +168 @@
           .getValue("matchBetweenFractions");
       
-      //TODO: remove later!!
-      boolean PRINT_POINTS = false;
-      
       
       // the percentage of the most intense features used to
@@ -737 +735 @@
           log.debug("Number of files in this fraction: " +nbrOfFiles);
           
+          TreeMap<Integer, TreeSet<Integer>> fileMatchMap = new TreeMap<Integer, TreeSet<Integer>>();
+          
+          for (int i=0;i<nbrOfFiles;i++){
+            TreeSet<Integer> tempSet = new TreeSet<Integer>();
+            tempSet.add(i);
+            fileMatchMap.put(i,tempSet);
+          }
+          
+          Set<Integer> maxSet = fileMatchMap.keySet();
+          Set<Integer> firstSubSet = fileMatchMap.get(0);
+          Set<Integer> secondSubSet = fileMatchMap.get(0);
+          int addedSetMatches = 0;
+          
           // new MATCHED_FILES_LIMIT needs to be computed for
           // every fraction-matching
@@ -757 +768 @@
           log.debug("Fraction based matched files limit: "
               + MATCHED_FILES_LIMIT);
-          boolean[] ignoreLimit = new boolean[nbrOfFiles];
-          
-          if(simList.size()<(nbrOfFiles*(nbrOfFiles-1))/2){
+          //boolean[] ignoreLimit = new boolean[nbrOfFiles];
+          
+          /*if(simList.size()<(nbrOfFiles*(nbrOfFiles-1))/2){
             int nRemovedPairs = (nbrOfFiles*(nbrOfFiles-1))/2 -simList.size(); 
             writer.println("Number of removed file-pairs from alignment: " +nRemovedPairs);
@@ -792 +803 @@
             
           }
-
+*/
 
           if (progress != null) {
@@ -821 +832 @@
           
           boolean breakAlignment = false;
+          boolean hasMatchedAllFiles = false;
+          boolean addedSetAlignment = false;
 
           int totNbrAlignments = simList.size();
@@ -858 +871 @@
                 .getSecondFile());
             
-            if (ignoreLimit[firstFileIndex] || ignoreLimit[secondFileIndex] || ((nbrMatchesPerFile[firstFileIndex]<currentMatch || nbrMatchesPerFile[secondFileIndex]<currentMatch) && nbrMatchesPerFile[firstFileIndex]<MATCHED_FILES_LIMIT+1 && nbrMatchesPerFile[secondFileIndex]<MATCHED_FILES_LIMIT+1)) {  
+            if (addedSetAlignment && firstSubSet.contains(fp.getFirstFile()) && secondSubSet.contains(fp.getSecondFile()) && addedSetMatches<MATCHED_FILES_LIMIT || ((nbrMatchesPerFile[firstFileIndex]<currentMatch || nbrMatchesPerFile[secondFileIndex]<currentMatch) && nbrMatchesPerFile[firstFileIndex]<MATCHED_FILES_LIMIT+1 && nbrMatchesPerFile[secondFileIndex]<MATCHED_FILES_LIMIT+1)) { 
+              
+              if (addedSetAlignment){
+                
+                addedSetMatches++;
+              }
+              
+              if (addedSetMatches==MATCHED_FILES_LIMIT){
+                writer.println("Subgroups have matched " +(addedSetMatches-1) +"times." );
+                addedSetAlignment = false;
+              }
               
               currentNbrALignments++;
@@ -887 +910 @@
                   + fp.getSimValue());
               writer.println();
+              
+              hasMatchedAllFiles=setMatchedFiles(fp.getFirstFile(), fp.getSecondFile(), fileMatchMap, nbrOfFiles, writer);
 
               // Alignment is to be performed, the matching
@@ -915 +940 @@
                   points, writer);
 
-              if(PRINT_POINTS == true){
-                writer.println("Points : "); for (Point
-               p:points){ writer.println(p.getX() +" "
-                +func.value(p.getX()) + " " +p.getY()); }
-                
-                writer.println();
-                writer.println("Eval points : "); for (Point
-                p:evalPoints){ writer.println(p.getX() +" "
-                +func.value(p.getX()) + " " +p.getY()); }
-               
-              }
               
               if (func == null) {
@@ -1059 +1073 @@
               itr.remove();
             }
-
-            if(!breakAlignment){
+            
+            if(!addedSetAlignment){
+
+              if(!breakAlignment){
+                for (int ii = 0; ii < nbrMatchesPerFile.length; ii++) {
+                  if (nbrMatchesPerFile[ii]< MATCHED_FILES_LIMIT) {
+                    break;
+                  }
+                  if(ii == nbrMatchesPerFile.length-1){
+                    breakAlignment=true;
+                  }
+                }
+
+              }
+            }
+
+            
+            if (breakAlignment) {
+              
+              if (hasMatchedAllFiles && !addedSetAlignment){
+                
+                simList.clear();
+                break;
+                
+              }else{
+                //there are subgroups of files that have not matched to each other
+                writer.println("There are subgroups in data.");
+                
+                addedSetAlignment = true;
+                
+                firstSubSet = fileMatchMap.get(0);
+                writer.println("First subgroup: " +firstSubSet.toString());
+                
+                maxSet.removeAll(firstSubSet);
+                secondSubSet = new TreeSet<Integer>(maxSet);
+                writer.println("Other subgroup: " +secondSubSet.toString());
+                
+                maxSet = fileMatchMap.keySet();
+                itr = simList.listIterator();
+                continue;
+                
+              }
+              
+              
+              
+            }
+            
+            if(!addedSetAlignment){
               for (int ii = 0; ii < nbrMatchesPerFile.length; ii++) {
-                if (nbrMatchesPerFile[ii]< MATCHED_FILES_LIMIT) {
+                if (nbrMatchesPerFile[ii]<currentMatch) {
                   break;
                 }
+
                 if(ii == nbrMatchesPerFile.length-1){
-                  breakAlignment=true;
+                  currentMatch++;
+                  itr = simList.listIterator();
                 }
-              }
-              
-            }
-            
-            
-            if (breakAlignment) {
-              
-              simList.clear();
-
-              break;
-            }
-
-            for (int ii = 0; ii < nbrMatchesPerFile.length; ii++) {
-              if (nbrMatchesPerFile[ii]<currentMatch) {
-                break;
-              }
-              
-              if(ii == nbrMatchesPerFile.length-1){
-                currentMatch++;
-                itr = simList.listIterator();
               }
             }
@@ -1423 +1465 @@
     }
 
+  }
+  
+  private boolean setMatchedFiles(int file1, int file2, TreeMap<Integer, TreeSet<Integer>> fileMatchMap, int nbrOfFiles, PrintWriter writer){
+    
+    Set<Integer> firstFileSet = fileMatchMap.get(file1);
+    
+      if(firstFileSet.size()==nbrOfFiles){
+        return true;
+      }
+      
+      if(firstFileSet.contains(file2)){
+        return false;
+      }
+    
+    Set<Integer> secondFileSet = fileMatchMap.get(file2);
+    
+    TreeSet<Integer> combinedSet = new TreeSet<Integer>();
+    combinedSet.addAll(firstFileSet);
+    combinedSet.addAll(secondFileSet);
+    
+    for(Integer key:firstFileSet){
+      fileMatchMap.put(key, combinedSet);
+    }
+    
+    for(Integer key:secondFileSet){
+      fileMatchMap.put(key, combinedSet);
+    }
+    
+    writer.println("Current map.");
+    for (int i=0; i<nbrOfFiles;i++){
+      writer.println(i +" " +fileMatchMap.get(i).toString());
+      
+    }
+    
+    if(combinedSet.size()==nbrOfFiles){
+      return true;
+    }else{
+      return false;
+    }
   }