Count overall occurrences when pooling

Author: mikessh

src/main/groovy/com/antigenomics/vdjtools/io/SampleWriter.groovy

@@ -297,7 +297,7 @@ public class SampleWriter {
     public void write(PooledSample pooledSample, String fileName) {
         def printWriter = getWriter(fileName)
 
-        printWriter.println(header + "\tincidence\tconvergence")
+        printWriter.println(header + "\tincidence\tconvergence\toccurrences")
 
         pooledSample.each { pooledClonotype ->
             printWriter.println(
@@ -310,7 +310,8 @@ public class SampleWriter {
                             pooledClonotype.clonotype."$it"
                     },
                      pooledClonotype.incidenceCount,
-                     pooledClonotype.diversity
+                     pooledClonotype.diversity,
+                     pooledClonotype.occurrences
                     ].flatten().join("\t"))
         }
 

src/main/java/com/antigenomics/vdjtools/model/CountMatrix.java

@@ -0,0 +1,9 @@
+package com.antigenomics.vdjtools.model;
+
+import com.antigenomics.vdjtools.sample.Clonotype;
+
+public interface CountMatrix {
+    void update(Clonotype clonotype);
+
+    long getCount(Clonotype clonotype);
+}

src/main/java/com/antigenomics/vdjtools/pool/StoringClonotypeAggregator.java

@@ -43,7 +43,7 @@
  */
 public class StoringClonotypeAggregator extends MaxClonotypeAggregator implements ClonotypeWrapper {
     private Clonotype clonotype;
-    private int convergence;
+    private int convergence, occurrences;
     private PooledSample parent;
     private final Set<ClonotypeKey> variants = new HashSet<>();
     private final ClonotypeKeyGen clonotypeKeyGen = new ClonotypeKeyGen();
@@ -52,6 +52,7 @@ protected StoringClonotypeAggregator(Clonotype clonotype, int sampleId) {
         super(clonotype, sampleId);
         this.clonotype = clonotype;
         this.convergence = 1;
+        this.occurrences = 1;
         this.variants.add(clonotypeKeyGen.generateKey(clonotype));
     }
 
@@ -63,6 +64,7 @@ protected boolean tryReplace(Clonotype clonotype, int sampleId) {
             convergence++;
             variants.add(clonotypeKey);
         }
+        occurrences++;
 
         if (super.tryReplace(clonotype, sampleId)) {
             this.clonotype = clonotype;
@@ -92,7 +94,8 @@ public void setParent(PooledSample parent) {
     }
 
     /**
-     * Gets the total number of convergent variants for this pooled clonotype.
+     * Gets the total number of convergent variants for this pooled clonotype, i.e.
+     * the total number of unique nucleotide variants in all samples.
      *
      * @return number of convergent variants.
      */
@@ -101,6 +104,17 @@ public int getDiversity() {
         return convergence;
     }
 
+    /**
+     * Gets the total number of clonotype occurrences - counting the total number of
+     * nucleotide variants in all samples. Differs from diversity/convergence as the
+     * same nucleotide variant found in two samples will be counted twice.
+     *
+     * @return number of occurrences
+     */
+    public int getOccurrences() {
+        return occurrences;
+    }
+
     /**
      * {@inheritDoc}
      */