Closes #591: don't replace reference on out-of-bounds variation

hgvs/variantmapper.py

@@ -23,7 +23,7 @@
 import hgvs.sequencevariant
 import hgvs.validator
 
-from hgvs.exceptions import HGVSUnsupportedOperationError, HGVSInvalidVariantError
+from hgvs.exceptions import HGVSDataNotAvailableError, HGVSUnsupportedOperationError, HGVSInvalidVariantError
 from hgvs.decorators.lru_cache import lru_cache
 from hgvs.enums import PrevalidationLevel
 from hgvs.utils.reftranscriptdata import RefTranscriptData
@@ -423,11 +423,8 @@ def _replace_reference(self, var):
         if var.type not in "cgmnr":
             raise HGVSUnsupportedOperationError("Can only update references for type c, g, m, n, r")
 
-        if var.posedit.edit.type == "ins":
-            # insertions have no reference sequence (zero-width), so return as-is
-            return var
-        if var.posedit.edit.type == "con":
-            # conversions have no reference sequence (zero-width), so return as-is
+        if var.posedit.edit.type in ("ins", "con"):
+            # these types have no reference sequence (zero-width), so return as-is
             return var
 
         pos = var.posedit.pos
@@ -443,7 +440,21 @@ def _replace_reference(self, var):
             pos = mapper.c_to_n(var.posedit.pos)
         else:
             pos = var.posedit.pos
-        seq = self.hdp.get_seq(var.ac, pos.start.base - 1, pos.end.base)
+
+        seq_start = pos.start.base - 1
+        seq_end = pos.end.base
+
+        # When strict_bounds is False and an error occurs, return
+        # variant as-is
+
+        try:
+            seq = self.hdp.get_seq(var.ac, seq_start, seq_end)
+        except HGVSDataNotAvailableError as e:
+            if (seq_start < 0 or len(seq) != seq_end - seq_start):
+                assert not hgvs.global_config.mapping.strict_bounds, f"{var}: Got out of bounds variant with strict_bounds enabled"
+                assert var.type in "cnr", f"Should not see out of bounds variant on type {var.type}"
+                _logger.info(f"{var}: variant outside sequence bounds; reference sequence can't be validated")
+                return var
 
         edit = var.posedit.edit
         if edit.ref != seq:

tests/issues/test_437.py

@@ -117,6 +117,23 @@ def test_enforce_strict_bounds(self):
         with pytest.raises(HGVSInvalidVariantError):
             e2_g = self.am37.n_to_g(self.e2_n)
 
+
+
+def test_invitae_examples(parser, am37):
+    tests = [
+        # ("NC_000009.11:g.35657741A>G", "NR_003051.3:n.*7T>C"), # Grammar doesn't support n.*
+        ("NC_000009.11:g.35658020C>T", "NR_003051.3:n.-5G>A"),
+        ]
+
+    hgvs.global_config.mapping.strict_bounds = False
+
+    for hgvs_g, hgvs_t in tests:
+        var_g = parser.parse(hgvs_g)
+        var_t = parser.parse(hgvs_t)
+        assert hgvs_t == str(am37.g_to_t(var_g, var_t.ac))
+
+    hgvs.global_config.mapping.strict_bounds = True
+
 
 
 def test_oob_dup(parser, am37):