Changeset 4068

Timestamp:

Dec 17, 2007, 9:02:44 AM (15 years ago)

Author:

Nicklas Nordborg

Message:

Fixes #880: Exclude reporters in Lowess

File:

• trunk/src/plugins/core/net/sf/basedb/plugins/LowessNormalization.java (modified) (18 diffs)

trunk/src/plugins/core/net/sf/basedb/plugins/LowessNormalization.java

@@ -33 +33 @@
 import net.sf.basedb.core.DynamicSpotQuery;
 import net.sf.basedb.core.FloatParameterType;
+import net.sf.basedb.core.Include;
 import net.sf.basedb.core.IntegerParameterType;
 import net.sf.basedb.core.Item;
+import net.sf.basedb.core.ItemParameterType;
+import net.sf.basedb.core.ItemQuery;
 import net.sf.basedb.core.Job;
 import net.sf.basedb.core.Permission;
 import net.sf.basedb.core.PluginParameter;
 import net.sf.basedb.core.ProgressReporter;
+import net.sf.basedb.core.ReporterList;
 import net.sf.basedb.core.RequestInformation;
 import net.sf.basedb.core.SpotBatcher;
@@ -55 +59 @@
 import net.sf.basedb.core.query.Expression;
 import net.sf.basedb.core.query.Expressions;
+import net.sf.basedb.core.query.Hql;
+import net.sf.basedb.core.query.JoinType;
 import net.sf.basedb.core.query.Orders;
 import net.sf.basedb.core.query.Restriction;
@@ -60 +66 @@
 import net.sf.basedb.core.query.Selects;
 import net.sf.basedb.core.query.SqlResult;
+import net.sf.basedb.core.query.WhenStatement;
 import net.sf.basedb.util.Values;
 
@@ -103 +110 @@
     "method addressing single and multiple slide systematic " +
     "variation. Nucleic Acids Res 2002, 30:e15.",
-    "2.5",
+    "2.6",
     "2006, Base 2 development team",
     null,
@@ -184 +191 @@
       permissions.add(new Permissions(Item.EXPERIMENT, null, EnumSet.of(Permission.USE)));
       permissions.add(new Permissions(Item.JOB, null, EnumSet.of(Permission.READ)));
+      permissions.add(new Permissions(Item.REPORTERLIST, null, EnumSet.of(Permission.READ)));
     }
     return permissions;
@@ -204 +212 @@
         float fitFraction = (Float) job.getValue(fitFractionParameter.getName());
         int iterations = (Integer) job.getValue(iterParameter.getName());
+        ReporterList excludeReporters = (ReporterList)job.getValue("excludeReporters");
         Job thisJob = getCurrentJob(dc);
         
-        BioAssaySet child = normalize(dc, source, thisJob, fitFraction, delta, iterations, blockGroupSize, progress);
+        BioAssaySet child = normalize(dc, source, thisJob, fitFraction, delta, iterations, 
+          blockGroupSize, excludeReporters, progress);
         child.setName(childName);
         child.setDescription(childDescription);
@@ -291 +301 @@
         storeValue(job, request, iterParameter);
         storeValue(job, request, blockGroupParameter);
+        storeValue(job, request, ri.getParameter("excludeReporters"));
         response.setDone("Job configuration complete", Job.ExecutionTime.SHORT);
       }
@@ -301 +312 @@
   // -------------------------------------------
 
+  /**
+    Normalise the source bioassay set using LOWESS normalization.
+    All reporters are included.
+    @since 2.5
+    @see #normalize(DbControl, BioAssaySet, Job, float, float, int, int, ReporterList, ProgressReporter)
+  */
+  public BioAssaySet normalize(DbControl dc, BioAssaySet source, Job job, float fitFraction, float delta, int iterations, int blockGroupSize, ProgressReporter progress)
+  {
+    return normalize(dc, source, job, fitFraction, delta, iterations, blockGroupSize, null, progress);
+  }
+  
+  
   /**
     Normalise the source bioassay set using LOWESS normalization.
@@ -311 +334 @@
     @param blockGroupSize The number of blocks to group and normalise at the same time,
       if <=0, all blocks on a bioassay are grouped to a single dataset
+    @param excludeReporters A reporter list containing reporters that should not be
+      used in when normalizing, or null to use all spots
     @param progress Progress reporter used by the caller to keep track of the progress. Null is allowed   
     @return The normalized bioassayset
     @since 2.5
   */
-  public BioAssaySet normalize(DbControl dc, BioAssaySet source, Job job, float fitFraction, float delta, int iterations, int blockGroupSize, ProgressReporter progress)
+  public BioAssaySet normalize(DbControl dc, BioAssaySet source, Job job, float fitFraction, float delta, int iterations, int blockGroupSize, ReporterList excludeReporters, ProgressReporter progress)
   {
     if (progress != null) progress.display(0, "Preparing to normalize...");
@@ -355 +380 @@
     if (progress != null) progress.display((int)(normalizedSpots / numSpots * 100), normalizedSpots + " spots normalized");
     
-    // Create query to retrieve spot data: COLUMN, POSITION, ch1, ch2, block
+    // If some reporters should be excluded, create an expression that:
+    // = 0 if spot's reporter is not present in list, = 1 if it is
+    Expression isPresent = Expressions.integer(0);
+    if (excludeReporters != null)
+    {
+      query.joinReporterList(excludeReporters, JoinType.LEFT);
+      isPresent = Expressions.caseWhen(isPresent, 
+        new WhenStatement(Dynamic.isPartOf(excludeReporters), Expressions.integer(1)));
+    }
+    
+    // Create query to retrieve spot data: COLUMN, POSITION, ch1, ch2, block, present
     // We use a parameter to restrict the query to return data for one bioassay at a time
+    // The 'exclude' value is 0 to not exclude and 1 to exclude a spot
     query.select(Dynamic.select(VirtualColumn.POSITION));
     query.select(Selects.expression(ch1, "ch1"));
     query.select(Selects.expression(ch2, "ch2"));
     query.select(Selects.expression(block, "block"));
+    query.select(Selects.expression(isPresent, "present"));
     query.restrict(bioAssayRestriction);
     query.order(Orders.asc(block));
@@ -384 +421 @@
         int ch2Index = it.getIndex("ch2");
         int blockIndex = it.getIndex("block");
+        int presentIndex = it.getIndex("present");
         
         // Copy bioassay data to SpotData objects
@@ -390 +428 @@
           SqlResult r = it.next();
           SpotData spot = new SpotData(r.getInt(positionIndex),
-            r.getFloat(ch1Index), r.getFloat(ch2Index), r.getInt(blockIndex));
+            r.getFloat(ch1Index), r.getFloat(ch2Index), r.getInt(blockIndex), 
+            r.getInt(presentIndex) == 1);
           data.add(spot);
         }
@@ -471 +510 @@
         parameters.add(iterParameter);
         parameters.add(blockGroupParameter);
+        PluginParameter<ReporterList> excludeReporters = 
+          getExcludeReportersParameters(dc);
+        if (excludeReporters != null) parameters.add(excludeReporters);
   
         configureJob = new RequestInformation
@@ -488 +530 @@
   }
 
+  private PluginParameter<ReporterList> getExcludeReportersParameters(DbControl dc)
+  {
+    // Load reporter types and initialise the reporterTypeParameter
+    ItemQuery<ReporterList> query = ReporterList.getQuery();
+    query.order(Orders.asc(Hql.property("name")));
+    query.include(Include.MINE, Include.SHARED, Include.IN_PROJECT);
+    List<ReporterList> reporterLists = new ArrayList<ReporterList>(query.list(dc));
+    if (reporterLists.size() == 0) return null;
+    
+    ItemParameterType<ReporterList> reporterListType = 
+      new ItemParameterType<ReporterList>(ReporterList.class, null, false, 1, reporterLists);
+    PluginParameter<ReporterList> reporterListParameter = new PluginParameter<ReporterList>(
+      "excludeReporters",
+      "Exclude reporters",
+      "Spots with reporters in the selected list will not be used in the lowess calculations. " +
+      "If no list is selected all spots are used.",
+      reporterListType);
+    return reporterListParameter;
+  }
+  
   private static List<Double> lowess(List<SpotData> data, double f, int iter, double delta)
   {
@@ -591 +653 @@
     for (int j = 0; j < data.size(); j++)
     {
-      double localW = w.get(j);
-      double localA = data.get(j).A;
-      double localM = data.get(j).M;
-      sumA += localA * localW;
-      sumM += localM * localW;
-      sumAA += localA * localA * localW;
-      sumAM += localA * localM * localW;
-      sumW += localW;
+      SpotData spot = data.get(j);
+      if (!spot.exclude)
+      {
+        double localW = w.get(j);
+        double localA = spot.A;
+        double localM = spot.M;
+        sumA += localA * localW;
+        sumM += localM * localW;
+        sumAA += localA * localA * localW;
+        sumAM += localA * localM * localW;
+        sumW += localW;
+      }
     }
     if (sumW <= 0)
@@ -634 +700 @@
     for (int i = 0; i < data.size(); i++)
     {
-      double A2 = data.get(i).A;
-      double distance = Math.abs(A1 - A2) * invRadius;
-      w.add((distance < 1 ? Math.pow(1D - Math.pow(Math.abs(distance), 3), 3) : 0) * wFit.get(i));
+      if (!data.get(i).exclude)
+      {
+        double A2 = data.get(i).A;
+        double distance = Math.abs(A1 - A2) * invRadius;
+        w.add((distance < 1 ? Math.pow(1D - Math.pow(Math.abs(distance), 3), 3) : 0) * wFit.get(i));
+      }
+      else
+      {
+        w.add(0.0);
+      }
     }
     return w;
@@ -650 +723 @@
     final double A;
     final int block;
+    final boolean exclude;
     static final double INV_LN2 = 1 / Math.log(2);
 
-    public SpotData(int position, float ch1, float ch2, int block)
+    public SpotData(int position, float ch1, float ch2, int block, boolean exclude)
     {
       this.position = position;
@@ -658 +732 @@
       this.ch2 = ch2;
       this.block = block;
+      this.exclude = exclude;
       this.M = Math.log(ch1 / ch2) * INV_LN2;
       this.A = Math.log10(Math.sqrt(ch1 * ch2));