From ede67b37af5da4e2588b215933ea442f5fc35c03 Mon Sep 17 00:00:00 2001
From: Tanisha <tanishasuhag127@gmail.com>
Date: Sun, 19 Apr 2026 01:32:16 +0530
Subject: [PATCH 1/2] fix: classify in-frame complex variants (MNP+INDEL) as
 CODON_CHANGE in veff.py

---
 malariagen_data/veff.py |  7 +++++--
 tests/test_veff.py      | 45 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 50 insertions(+), 2 deletions(-)

diff --git a/malariagen_data/veff.py b/malariagen_data/veff.py
index f3699792c..3b08825d6 100644
--- a/malariagen_data/veff.py
+++ b/malariagen_data/veff.py
@@ -444,9 +444,12 @@ def _get_within_cds_effect(ann, base_effect, cds, cdss):
             effect = base_effect._replace(effect="CODON_CHANGE", impact="MODERATE")
 
         else:
-            # TODO in-frame complex variation (MNP + INDEL)
+            # In-frame complex variation (MNP + INDEL combined), where the
+            # net length change is a multiple of 3, so no frameshift occurs.
+            # Classified as CODON_CHANGE with MODERATE impact, consistent
+            # with how pure MNPs are handled above.
             effect = base_effect._replace(
-                effect="TODO in-frame complex variation (MNP + INDEL)", impact="UNKNOWN"
+                effect="CODON_CHANGE", impact="MODERATE"
             )
 
     return effect
diff --git a/tests/test_veff.py b/tests/test_veff.py
index dfe036979..48776d8a0 100644
--- a/tests/test_veff.py
+++ b/tests/test_veff.py
@@ -683,3 +683,48 @@ def test_isolation(self):
         ann1.clear_genome_cache()
         assert ann1._load_genome_seq.cache_info().currsize == 0
         assert ann2._load_genome_seq.cache_info().currsize == 1
+
+
+# ── TestInFrameComplexVariant ─────────────────────────────────────────────────
+
+
+class TestInFrameComplexVariant:
+    """
+    Verify that in-frame complex variants (MNP + INDEL combined, where the
+    net length change is a multiple of 3) are classified as CODON_CHANGE
+    with MODERATE impact — not the old TODO string with UNKNOWN impact.
+
+    The forward-strand CDS is: ATG|GCC|TTA|CAG|TGA (pos 4-18)
+    We use pos 7 (start of GCC codon) as our test position.
+
+    Example variant:
+        ref = "GCC"   (3 bp, the whole Ala codon at pos 7-9)
+        alt = "GCCATG" (6 bp — same start codon, plus a new Met codon inserted)
+        net length change = 6 - 3 = 3, which is a multiple of 3 → in-frame
+        len(ref) = 3 > 1, len(alt) = 6 > 1 → not a simple insertion or deletion
+        len(ref) != len(alt) → not a pure MNP
+        This hits the previously-unhandled else branch.
+    """
+
+    def setup_method(self):
+        self.ann = Annotator(make_genome(("chr1", FWD_SEQ)), FEATURES_BASIC_FWD.copy())
+
+    def test_inframe_complex_classified_as_codon_change(self):
+        # ref="GCC" (3 bp), alt="GCCATG" (6 bp): net change = +3, in-frame
+        row = _run(self.ann, "tx1", 7, "GCC", "GCCATG")
+        assert row["effect"] == "CODON_CHANGE"
+        assert row["impact"] == "MODERATE"
+
+    def test_inframe_complex_not_todo_string(self):
+        # Make sure the old TODO string is never returned
+        row = _run(self.ann, "tx1", 7, "GCC", "GCCATG")
+        assert "TODO" not in str(row["effect"])
+        assert row["impact"] != "UNKNOWN"
+
+    def test_inframe_complex_deletion_classified_as_codon_change(self):
+        # ref="GCCTTA" (6 bp), alt="GCC" (3 bp): net change = -3, in-frame
+        # len(ref)=6 > 1, len(alt)=3 > 1 → not a simple deletion
+        row = _run(self.ann, "tx1", 7, "GCCTTA", "GCC")
+        assert row["effect"] == "CODON_CHANGE"
+        assert row["impact"] == "MODERATE"
+        assert "TODO" not in str(row["effect"])
\ No newline at end of file

From 5342a95f3d27962a52dacf20ccca434b3f1e262e Mon Sep 17 00:00:00 2001
From: Tanisha <tanishasuhag127@gmail.com>
Date: Sun, 19 Apr 2026 22:28:54 +0530
Subject: [PATCH 2/2] style: fix ruff-format and end-of-file issues

---
 malariagen_data/veff.py | 4 +---
 tests/test_veff.py      | 2 +-
 2 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/malariagen_data/veff.py b/malariagen_data/veff.py
index 3b08825d6..32c5e16c2 100644
--- a/malariagen_data/veff.py
+++ b/malariagen_data/veff.py
@@ -448,9 +448,7 @@ def _get_within_cds_effect(ann, base_effect, cds, cdss):
             # net length change is a multiple of 3, so no frameshift occurs.
             # Classified as CODON_CHANGE with MODERATE impact, consistent
             # with how pure MNPs are handled above.
-            effect = base_effect._replace(
-                effect="CODON_CHANGE", impact="MODERATE"
-            )
+            effect = base_effect._replace(effect="CODON_CHANGE", impact="MODERATE")
 
     return effect
 
diff --git a/tests/test_veff.py b/tests/test_veff.py
index 48776d8a0..d4b2b753f 100644
--- a/tests/test_veff.py
+++ b/tests/test_veff.py
@@ -727,4 +727,4 @@ def test_inframe_complex_deletion_classified_as_codon_change(self):
         row = _run(self.ann, "tx1", 7, "GCCTTA", "GCC")
         assert row["effect"] == "CODON_CHANGE"
         assert row["impact"] == "MODERATE"
-        assert "TODO" not in str(row["effect"])
\ No newline at end of file
+        assert "TODO" not in str(row["effect"])