Changeset 729

Timestamp:

Jul 21, 2008, 4:42:06 PM (15 years ago)

Author:

Martin Svensson

Message:

References #117 Almost done with implementing the normalization algorithm for this plugin

Location:

plugins/base2/net.sf.basedb.normalizations/trunk/src/net/sf/basedb/plugins

Files:

• AbstractNormalizationPlugin.java (modified) (3 diffs)
• AverageNormalization.java (modified) (9 diffs)

plugins/base2/net.sf.basedb.normalizations/trunk/src/net/sf/basedb/plugins/AbstractNormalizationPlugin.java

@@ -49 +49 @@
 {
 
-  private static final String REFERENCE_GROUP = "reference_group";
+  protected static final String REFERENCE_GROUP = "reference_group";
   
   private static final Set<Permissions> permissions = new HashSet<Permissions>();
@@ -73 +73 @@
   }
   
+  /**
+    Gets the plug-in parameter that holds the reference group.
+      @param label
+      @param description
+      @return PluginParameter for a BioAssay
+   */
   protected PluginParameter<BioAssay> getReferenceGroupParameter(String label, String description)
   {
@@ -80 +86 @@
       if (description == null) 
       {
-        description = "Select bio assays to be used as a reference group. " +
+        description = "Select one or several bio assays to be used as a reference group. " +
             "The reference group is used to calculate a virtual array. ";
       }

plugins/base2/net.sf.basedb.normalizations/trunk/src/net/sf/basedb/plugins/AverageNormalization.java

@@ -30 +30 @@
 import net.sf.basedb.core.BioAssaySet;
 import net.sf.basedb.core.DbControl;
+import net.sf.basedb.core.DynamicQuery;
+import net.sf.basedb.core.DynamicResultIterator;
+import net.sf.basedb.core.DynamicSpotQuery;
+import net.sf.basedb.core.Job;
 import net.sf.basedb.core.PluginParameter;
 import net.sf.basedb.core.ProgressReporter;
 import net.sf.basedb.core.RequestInformation;
+import net.sf.basedb.core.SpotBatcher;
+import net.sf.basedb.core.Transformation;
+import net.sf.basedb.core.Type;
+import net.sf.basedb.core.VirtualColumn;
 import net.sf.basedb.core.plugin.About;
 import net.sf.basedb.core.plugin.AboutImpl;
@@ -39 +47 @@
 import net.sf.basedb.core.plugin.Request;
 import net.sf.basedb.core.plugin.Response;
+import net.sf.basedb.core.query.Aggregations;
+import net.sf.basedb.core.query.Dynamic;
+import net.sf.basedb.core.query.Expression;
+import net.sf.basedb.core.query.Expressions;
+import net.sf.basedb.core.query.Orders;
+import net.sf.basedb.core.query.Restriction;
+import net.sf.basedb.core.query.Restrictions;
+import net.sf.basedb.core.query.Selects;
+import net.sf.basedb.core.query.SqlResult;
 import net.sf.basedb.core.signal.SignalHandler;
 import net.sf.basedb.core.signal.SignalReceivedException;
@@ -46 +63 @@
 import net.sf.basedb.util.Values;
 
+import java.sql.SQLException;
 import java.util.ArrayList;
 import java.util.Arrays;
@@ -72 +90 @@
   private RequestInformation configureJob = null;
   
+  /**
+    Expressions used to get data
+  */
+  private Expression ch1 = Dynamic.column(VirtualColumn.channel(1));
+  
+  /**
+    Create restriction: column = :bioAssayColumn
+  */
+  Restriction bioAssayRestriction = Restrictions.eq(
+      Dynamic.column(VirtualColumn.COLUMN),
+      Expressions.parameter("bioAssayColumn")
+    );
   
   private final About about = new AboutImpl
@@ -124 +154 @@
         // Child name, description and transformation
         storeValue(job, request, ri.getParameter(CHILD_NAME));
-        storeValue(job, request, ri.getParameter(CHILD_DESCRIPTION));
-        
-        //TODO Store Average Normalization specific parameters
+        storeValue(job, request, ri.getParameter(CHILD_DESCRIPTION));       
+        storeValue(job, request, ri.getParameter(REFERENCE_GROUP));
+        //TODO More normalization specific parameters??
+        
+        response.setDone("Job configuration completed", Job.ExecutionTime.MEDIUM);
       }
       catch (SignalReceivedException signalException)
@@ -181 +213 @@
         String childName = Values.getString((String)job.getValue(CHILD_NAME), source.getName());
         String childDescription = (String)job.getValue(CHILD_DESCRIPTION);
-        
-        List<BioAssay> referenceGroup = getReferenceGroup(dc);
-        //TODO Implement things specific for the algortihm.
+        Job thisJob = getCurrentJob(dc);
+        
+        List<BioAssay> refGroup = getReferenceGroup(dc);
+        
+        //Normalize the data
+        BioAssaySet child = normalize(dc, source, thisJob, refGroup, progress);
+        
+        //Finish up
+        child.setName(childName);
+        child.setDescription(childDescription);
+        dc.commit();
+        int normalizedSpots = child.getNumSpots();
+        if (progress != null) progress.display(100, normalizedSpots + " spots normalized\n");
+        response.setDone(normalizedSpots + " spots normalized, " + (source.getNumSpots() - normalizedSpots) + " spots removed");
       }
       catch (Throwable ex)
@@ -198 +241 @@
    ---------------------------------------------------------
   */
+  
+  public BioAssaySet normalize(DbControl dc, BioAssaySet source, Job job, List<BioAssay> refGroup, ProgressReporter progress)
+  {
+    if (progress != null) progress.display(0, "Preparing normalization...");
+    
+    // Create transformation and prepare for the normalization
+    Transformation t = source.newTransformation(job);
+    t.setName(about.getName());
+    dc.saveItem(t);
+    
+    BioAssaySet child = t.newProduct(null, "new", true);    
+    dc.saveItem(child);
+    
+    SpotBatcher spotBatcher = child.getSpotBatcher();
+
+    float scaleFactor = getScaleFactor(dc, refGroup, source, progress);
+    // Get the number of spots to be normalized
+    DynamicSpotQuery query = source.getSpotData();
+    long totalNumSpots = query.count(dc);
+    query.reset();
+    long normalizedSpots = 0;
+    
+    // Restrictions for the ch1's value
+    query.select(Dynamic.select(VirtualColumn.POSITION));
+    query.select(Selects.expression(ch1, "ch1"));
+    query.restrict(bioAssayRestriction);
+    
+    if (progress != null)
+      progress.display((int)(normalizedSpots / totalNumSpots * 100),
+                       normalizedSpots + " spots normalized");
+    
+    // Normalize data
+    List<BioAssay> assays = source.getBioAssays().list(dc);
+    try
+    {
+      for (BioAssay ba : assays)
+      {
+        checkInterrupted();
+        
+        short bioAssayColumn = ba.getDataCubeColumnNo();
+        query.setParameter("bioAssayColumn", (int)bioAssayColumn, Type.INT);
+        DynamicResultIterator it = query.iterate(dc);
+        int positionIndex = it.getIndex(VirtualColumn.POSITION.getName());
+        int ch1Index = it.getIndex("ch1");
+        
+        while (it.hasNext())
+        {
+          SqlResult r = it.next();
+          float intensity = r.getFloat(ch1Index);
+          int position = r.getInt(positionIndex);         
+          
+          spotBatcher.insert(bioAssayColumn, position, (intensity * scaleFactor));
+          ++normalizedSpots;
+        }
+        it.close();
+        spotBatcher.flush();
+        if (progress != null)
+          progress.display((int)((normalizedSpots * 100L) / totalNumSpots), 
+              normalizedSpots + " spots normalized");
+      }
+    }
+    catch (SQLException ex)
+    {
+      throw new BaseException(ex);
+    }
+    return child;
+  }
+  
   
   private RequestInformation getConfiguredJobParameters()
@@ -217 +328 @@
         //Reference group
         parameters.add(getReferenceGroupParameter(null, null));
-        //TODO Configure Average Normalization specific parameters.
+        //TODO More normalization specific parameters???
       }
     }
@@ -227 +338 @@
     return configureJob;
   }
+  
+  private float getScaleFactor(DbControl dc, List<BioAssay> refGroup, BioAssaySet source, ProgressReporter progress)
+  { 
+    float averageVirtualArray;
+    float averageWholeGroup;
+    
+    DynamicSpotQuery query = source.getSpotData();
+    query.selectPermanent(Selects.expression(Aggregations.sum(ch1), "sumCh1"));
+    query.restrictPermanent(bioAssayRestriction);
+    
+    //Get average intensity for the bio assays in reference group.   
+    float totSum = 0;
+    long numOfSpots = 0;
+    try
+    {
+      for (BioAssay ba : refGroup)
+      {
+        numOfSpots = numOfSpots + ba.getNumSpots();
+        short bioAssayColumn = ba.getDataCubeColumnNo();
+        query.setParameter("bioAssayColumn", (int)bioAssayColumn, Type.INT);
+        DynamicResultIterator it = query.iterate(dc);
+        int sumCh1Index = it.getIndex("sumCh1");
+        SqlResult result = it.next();
+        totSum = totSum + result.getFloat(sumCh1Index);
+        it.close();
+      }     
+    }
+    catch (SQLException e)
+    {
+      throw new BaseException(e);
+    }   
+    averageVirtualArray = totSum / numOfSpots;
+    query.reset();
+    
+    // Get average intensity for the whole group
+    List<BioAssay> assays = source.getBioAssays().list(dc);
+    totSum = 0;
+    numOfSpots = 0;
+    try
+    {
+      for (BioAssay ba : assays)
+      {
+        numOfSpots = numOfSpots + ba.getNumSpots();
+        short bioAssayColumn = ba.getDataCubeColumnNo();
+        query.setParameter("bioAssayColumn", (int)bioAssayColumn, Type.INT);
+        DynamicResultIterator it = query.iterate(dc);
+        int sumCh1Index = it.getIndex("sumCh1");
+        SqlResult result = it.next();
+        totSum = totSum + result.getFloat(sumCh1Index);
+        it.close();
+      }
+    }
+    catch (SQLException e)
+    {
+      throw new BaseException(e);
+    }
+    averageWholeGroup = totSum / numOfSpots;
+    
+    
+    float scaleFactor = averageVirtualArray / averageWholeGroup;
+    
+    return scaleFactor;
+  }
 }