Merge pull request #73 from GavinHuttley/main

Fix newick output

src/diverse_seq/cli.py

@@ -457,7 +457,6 @@ def ctree(
     hide_progress: bool,
 ):
     """Quickly compute a cluster tree based on kmers for a collection of sequences."""
-
     if seqfile.suffix != ".dvseqs":
         dvs_util.print_colour(
             "Sequence data needs to be preprocessed, use 'dvs prep'",
@@ -492,7 +491,6 @@ def ctree(
         seqs[name] = arr2str_app(record.read())  # pylint: disable=not-callable
 
     seqs = make_unaligned_seqs(seqs, moltype=moltype)
-
     app = dvs_cluster.dvs_par_ctree(
         k=k,
         sketch_size=sketch_size,
@@ -504,6 +502,9 @@ def ctree(
         show_progress=not hide_progress,
     )
     tree = app(seqs)  # pylint: disable=not-callable
+    if not tree:
+        dvs_util.print_colour(tree, "red")
+        sys.exit(1)
     tree.write(outpath)
 
 

src/diverse_seq/cluster.py

@@ -231,11 +231,10 @@ def make_cluster_tree(
             tree_dict.pop(right_index),
         )
         node_index += 1
-
-    tree = make_tree(str(tree_dict[node_index - 1]))
-
+    # use string representation and then remove quotes
+    treestring = str(tree_dict[node_index - 1]).replace("'", "")
+    tree = make_tree(treestring=treestring, underscore_unmunge=True)
     progress.update(tree_task, completed=1, total=1)
-
     return tree
 
 
@@ -297,7 +296,6 @@ def __init__(
             mash_canonical_kmers=mash_canonical_kmers,
             show_progress=show_progress,
         )
-
         if parallel:
             if max_workers is None:
                 max_workers = multiprocessing.cpu_count()

src/diverse_seq/distance.py

@@ -159,7 +159,6 @@ def mash_distances(
     numpy.ndarray
         Pairwise mash distances between sequences.
     """
-
     if progress is None:
         progress = Progress(disable=True)
 

tests/test_cli.py

@@ -253,9 +253,7 @@ def test_ctree(
 ):
     outpath = tmp_dir / "out.tre"
 
-    args = (
-        f"-s {processed_seq_path} -o {outpath} -d {distance} -k {k} -np {max_workers}"
-    )
+    args = f"-s {processed_seq_path} -o {outpath} -d {distance} -k {k} -np {max_workers} -hp"
     if sketch_size is not None:
         args += f" --sketch-size {sketch_size}"
     args = args.split()