Merge branch 'biocommons:main' into issue_178_abc_EDit

biocommons · Mar 21, 2024 · 76e5f01 · 76e5f01
2 parents 0734cfa + 3c6c265
commit 76e5f01
Show file tree

Hide file tree

Showing 3 changed files with 27 additions and 12 deletions.
diff --git a/src/hgvs/extras/babelfish.py b/src/hgvs/extras/babelfish.py
@@ -74,16 +74,24 @@ def hgvs_to_vcf(self, var_g):
         return chrom, start_i + 1, ref, alt, typ
 
     def vcf_to_g_hgvs(self, chrom, position, ref, alt):
+        # VCF spec https://samtools.github.io/hts-specs/VCFv4.1.pdf
+        # says for REF/ALT "Each base must be one of A,C,G,T,N (case insensitive)"
+        ref = ref.upper()
+        alt = alt.upper()
+
         ac = self.name_to_ac_map[chrom]
 
-        # Strip common prefix
-        if len(alt) > 1 and len(ref) > 1:
-            pfx = os.path.commonprefix([ref, alt])
-            lp = len(pfx)
-            if lp > 0:
-                ref = ref[lp:]
-                alt = alt[lp:]
-                position += lp
+        if ref != alt:
+            # Strip common prefix
+            if len(alt) > 1 and len(ref) > 1:
+                pfx = os.path.commonprefix([ref, alt])
+                lp = len(pfx)
+                if lp > 0:
+                    ref = ref[lp:]
+                    alt = alt[lp:]
+                    position += lp
+            elif alt == ".":
+                alt = ref
 
         if ref == "":  # Insert
             # Insert uses coordinates around the insert point.
@@ -93,9 +101,6 @@ def vcf_to_g_hgvs(self, chrom, position, ref, alt):
             start = position
             end = position + len(ref) - 1
 
-        if alt == ".":
-            alt = ref
-
         var_g = SequenceVariant(
             ac=ac,
             type="g",

diff --git a/tests/data/cache-py3.hdp b/tests/data/cache-py3.hdp
diff --git a/tests/test_hgvs_extras_babelfish.py b/tests/test_hgvs_extras_babelfish.py
@@ -8,7 +8,17 @@
         "NC_000006.12:g.49949407=",
         [],
         ("6", 49949407, "A", ".", "identity"),
-        [("6", 49949407, "A", "A", "identity")],
+        [("6", 49949407, "A", "A", "identity"),
+         # Test case insensitivity
+         ("6", 49949407, "A", "a", "identity"),
+         ("6", 49949407, "a", "A", "identity"),]
+    ),
+    # Test multi-base identity
+    (
+        "NC_000006.12:g.49949407_49949408=",
+        [],
+        ("6", 49949407, "AA", ".", "identity"),
+        [("6", 49949407, "AA", "AA", "identity")],
     ),
     # snv
     (