add some tests for minus strand variant calling

countess/utils/variant.py

@@ -212,7 +212,6 @@ def find_variant_dna(ref_seq: str, var_seq: str, offset: int = 0) -> Iterable[st
 
     >>> list(find_variant_dna("AGAAGTAGAGG", "ATAAGAAGAGG", -200))
     ['198G>T', '194T>A']
-
     """
 
     ref_seq = ref_seq.strip().upper()
@@ -524,6 +523,13 @@ def find_variant_string(
     >>> find_variant_string("p.", "ATGGTTGGTTCA", "ATGGCTGCTTCA")
     'p.Val2_Gly3delinsAlaAla'
 
+    MINUS STRAND
+
+    this example is actually comparing TGTAATC and TCTGAAC ...
+
+    >>> find_variant_string("g.", "GATTACA", "GTTCAGA", minus_strand=True)
+    'g.[2G>C;3_4insG;6del]'
+
     CHECK FOR INVALID INPUTS
 
     >>> find_variant_string("x.", "CAT", "CAT")

tests/plugins/test_variant.py

@@ -69,3 +69,42 @@ def test_variant_ref_offset():
     assert output[0]["out"] == "g.1A>T"
     assert output[1]["out"] == "g.[17A>T;19A>T]"
     assert output[2]["out"] == "g.[102G>T;106T>A]"
+
+def test_variant_ref_offset_minus():
+    """check that the reverse-complement behaviour works on the minus strand."""
+    # genes on the minus strand are reverse-complemented, so what we're actually
+    # comparing is the reverse-complemented sequences:
+    #
+    #      00000000011
+    # num  12345678901
+    # ref  CCTCTACTTCT
+    # seq1 CCTCTACTTCA => 11T>A
+    # seq2 CCACAACTTCT => 3T>A;5T>A
+    # seq3 CCTCTTCTTAT => 6A>T;10C>A
+    #
+    # plus the offset
+
+    input_df = pd.DataFrame(
+        [
+            {"seq": "TGAAGTAGAGG" },
+            {"seq": "AGAAGTTGTGG" },
+            {"seq": "ATAAGAAGAGG" },
+        ]
+    )
+
+    plugin = VariantPlugin()
+    plugin.set_parameter("column", "seq")
+    plugin.set_parameter("reference", "AGAAGTAGAGG")
+    plugin.set_parameter("outputs.0.offset", "1000")
+    plugin.set_parameter("outputs.0.seq_type", "g-")
+    plugin.set_parameter("outputs.0.output", "out")
+
+    plugin.prepare(["test"], None)
+
+    output_df = plugin.process_dataframe(input_df)
+
+    output = output_df.to_records()
+
+    assert output[0]["out"] == "g.1011T>A"
+    assert output[1]["out"] == "g.[1003T>A;1005T>A]"
+    assert output[2]["out"] == "g.[1006A>T;1010C>A]"