replace param min_samples with dynamic calculation

workflow/scripts/read_cluster.py

@@ -213,7 +213,6 @@ def parse_args():
     cluster_parser = subparsers.add_parser('cluster', help="Cluster reads by distance.")
     cluster_parser.add_argument("sequences", type=Path)
     cluster_parser.add_argument("distances", type=Path)
-    cluster_parser.add_argument("--min_samples", type=int, default=5)
     cluster_parser.add_argument("--eps", type=float, default=0.3)
     cluster_parser.add_argument("--output", "-o", type=Path, default=None)
 
@@ -229,13 +228,15 @@ def parse_args():
 if __name__ == '__main__':
     args = parse_args()
     seq_coll = SequenceCollection.from_fasta(args.sequences)
+    n_reads = seq_coll.n_reads()
     n_excluded = len(seq_coll.get_labels())
     if args.command == "cluster":
         # Inputs are: distances, output, eps, min_samples
         # produces a file (output) that contains the cluster labels for all given sequences
         if args.distances is not None:
             matrix = read_dist_matrix(args.distances)
-            clustering = cluster_on_distances(matrix, args.eps, args.min_samples, n_excluded)
+            clustering = cluster_on_distances(matrix, args.eps,
+                    min_samples=min(n_reads//3+1, 5), exclude_last_n=n_excluded)
             write_clustering(clustering, args.output)
         elif args.sequences is None:
             print("You have to either pass pairwise distances or sequences to be clustered.")