From 84d385c9e87e533f3f588b8d24cc75086ac2ad8d Mon Sep 17 00:00:00 2001
From: Gregor Sturm <mail@gregor-sturm.de>
Date: Thu, 11 Sep 2025 19:57:09 +0200
Subject: [PATCH] Add more tests for clonotype clustering

---
 src/scirpy/tests/test_clonotypes.py | 37 ++++++++++++++++++++++++++++-
 src/scirpy/tl/_clonotypes.py        |  3 ++-
 2 files changed, 38 insertions(+), 2 deletions(-)

diff --git a/src/scirpy/tests/test_clonotypes.py b/src/scirpy/tests/test_clonotypes.py
index 081abff00..da34eaa63 100644
--- a/src/scirpy/tests/test_clonotypes.py
+++ b/src/scirpy/tests/test_clonotypes.py
@@ -163,6 +163,14 @@ def test_clonotypes_end_to_end1(adata_define_clonotypes):
             [0, 0, 0, 1, 0, np.nan, 0, 0, 0, 0, 2],
             [8, 8, 8, 1, 8, np.nan, 8, 8, 8, 8, 1],
         ),
+        (
+            "VDJ",
+            "primary_only",
+            False,
+            None,
+            [0, 0, 0, 1, 0, np.nan, 0, 2, 3, 3, np.nan],
+            [5, 5, 5, 1, 5, np.nan, 5, 1, 2, 2, np.nan],
+        ),
         (
             "VDJ",
             "primary_only",
@@ -171,6 +179,33 @@ def test_clonotypes_end_to_end1(adata_define_clonotypes):
             [0, 0, 0, 1, 0, np.nan, 0, 2, 3, 4, 5],
             [5, 5, 5, 1, 5, np.nan, 5, 1, 1, 1, 1],
         ),
+        (
+            "VJ",
+            "primary_only",
+            False,
+            None,
+            [0, 0, 1, 1, 0, np.nan, 0, 0, 0, np.nan, np.nan],
+            [6, 6, 2, 2, 6, np.nan, 6, 6, 6, np.nan, np.nan],
+        ),
+        (
+            "VJ",
+            "all",
+            False,
+            None,
+            [0, 0, 1, 1, 2, np.nan, 2, 0, 2, np.nan, np.nan],
+            [3, 3, 2, 2, 3, np.nan, 3, 3, 3, np.nan, np.nan],
+        ),
+        (
+            "VJ",
+            "all",
+            True,
+            None,
+            # TODO: the last two cells get assigned clonotype cluster 5 because they have a v gene, but no junction_aa sequence
+            # Do we want that? Or should it be np.nan if no sequence exists? Typically this would
+            # be filtered out before.
+            [0, 1, 2, 3, 4, np.nan, 4, 1, 4, np.nan, np.nan],
+            [1, 1, 2, 1, 3, np.nan, 3, 2, 3, np.nan, np.nan],
+        ),
         # v gene and receptor type
         (
             "any",
@@ -206,7 +241,7 @@ def test_clonotype_clusters_end_to_end(
         same_v_gene=same_v_gene,
     )  # type: ignore
     print(clonotypes)
-    npt.assert_equal(list(clonotypes.values), [str(x) if not np.isnan(x) else x for x in expected])
+    assert list(clonotypes.values) == [str(x) if not np.isnan(x) else x for x in expected]
     npt.assert_almost_equal(clonotype_size.values, expected_size)
 
 
diff --git a/src/scirpy/tl/_clonotypes.py b/src/scirpy/tl/_clonotypes.py
index 528a67f93..7f8e67b75 100644
--- a/src/scirpy/tl/_clonotypes.py
+++ b/src/scirpy/tl/_clonotypes.py
@@ -125,7 +125,7 @@
 
 def _validate_parameters(
     params: DataHandler,
-    reference: DataHandler,
+    reference: DataHandler | None,
     receptor_arms,
     dual_ir,
     within_group,
@@ -138,6 +138,7 @@ def _validate_parameters(
 
     def _get_db_name():
         try:
+            assert reference is not None
             return reference.adata.uns["DB"]["name"]
         except KeyError:
             raise ValueError(