Bugfix: MAX_CLUSTERS capping

When reach MAX_CLUSTERS, use that number of clusters in output, rather than put each sequence in its own cluster.
iqbal-lab-org · Oct 26, 2020 · a496215 · a496215
1 parent f99c35d
commit a496215
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 3 deletions.
diff --git a/make_prg/from_msa/cluster_sequences.py b/make_prg/from_msa/cluster_sequences.py
@@ -175,8 +175,8 @@ def kmeans_cluster_seqs_in_interval(
 
         while cluster_further(seqclustering):
             num_clusters += 1
-            if num_clusters >= MAX_CLUSTERS:
-                num_clusters = num_sequences
+            if num_clusters > MAX_CLUSTERS:
+                break
             if num_clusters == num_sequences:
                 break
             start = time.time()

diff --git a/tests/from_msa/test_cluster_sequences.py b/tests/from_msa/test_cluster_sequences.py
@@ -18,6 +18,7 @@
     get_distances,
     DISTANCE_THRESHOLD,
     LENGTH_THRESHOLD,
+    MAX_CLUSTERS,
     get_one_ref_like_threshold_distance,
     sequences_are_one_reference_like,
     cluster_further,
@@ -388,7 +389,7 @@ def test_GivenManyVeryDifferentSequences_EachSeqInOwnCluster(self):
         all_5mers = list(map("".join, product(standard_bases, repeat=5)))
         alignment = make_alignment(all_5mers)
         result = kmeans_cluster_seqs_in_interval([0, 5], alignment, 5)
-        self.assertEqual(len(result), len(all_5mers))
+        self.assertEqual(len(result), MAX_CLUSTERS)
 
 
 class TestMergeClusters(TestCase):