WindowsStorage: get the germlines of the most abundant clones

algo/core/windows.cpp

@@ -73,6 +73,21 @@ list<Sequence> WindowsStorage::getSample(junction window, size_t nb_sampled,
   return SequenceSampler(reads).getLongest(nb_sampled, nb_buckets);
 }
 
+set<Germline *> WindowsStorage::getTopGermlines(size_t top, size_t min_reads) {
+  assert(sort_all_windows.size() == seqs_by_window.size());
+
+  set<Germline *> top_germlines;
+  size_t count = 0;
+
+  for (list<pair <junction, int> >::const_iterator it = sort_all_windows.begin();
+       it != sort_all_windows.end() && count < top && (size_t)it->second >= min_reads;
+       ++it, ++count) {
+    top_germlines.insert(getGermline(it->first));
+  }
+
+  return top_germlines;
+}
+
 size_t WindowsStorage::size() {
   return seqs_by_window.size();
 }

algo/core/windows.h

@@ -11,6 +11,7 @@
 
 #include <iostream>
 #include <map>
+#include <set>
 #include <utility>
 #include <string>
 #include "fasta.h"
@@ -103,6 +104,16 @@ class WindowsStorage {
    */
   string getLabel(junction window);
 
+  /**
+   * @pre sort() must have been called.
+   * @param top: Only the germlines of the top most abundant windows will
+   *             be considered
+   * @param min_reads: (optional) minimal number (inclusive) of reads the window
+   *                   must be supported by.
+   * @return a set of the most abundant germlines.
+   */
+  set<Germline *> getTopGermlines(size_t top, size_t min_reads=1);
+
   /**
    * Only keep windows that are interesting.  Those windows are windows
    * supported by at least min_reads_window reads as well as windows that are