Merge pull request #92 from bcgsc/bugfix/DEVSU-2793-prune-nostmt-matches-from-keyalts

exclude partial-only matched kbvars from key alts

Author: elewis2

pori_python/ipr/ipr.py

@@ -269,11 +269,22 @@ def select_expression_plots(
 
 
 def create_key_alterations(
-    kb_matches: List[Hashabledict], all_variants: Sequence[IprVariant]
+    kb_matches: List[Hashabledict],
+    all_variants: Sequence[IprVariant],
+    included_kb_matches: List[KbVariantMatch],
 ) -> Tuple[List[Dict], Dict]:
     """Create the list of significant variants matched by the KB.
 
     This list of matches is also used to create the variant counts.
+
+    kb_matches: the full list of matched kb objects found for the reported variants
+    all_variants: the full list of all reported variants, matched or unmatched
+    included_kb_matches: the list of kb_variant ids to be allowed in the key alterations table;
+        this is all kb_variants if partially matched statements are allowed, or
+        the subset of kb_variants that are conditions for at least one
+        fully satisfied statement condition set, if partially matched statements
+        are not allowed (ie, kb_variants that are not part of any fully satisfied
+        statement condition set are excluded)
     """
     alterations = []
     type_mapping = {
@@ -284,7 +295,12 @@ def create_key_alterations(
     }
     counts: Dict[str, Set] = {v: set() for v in type_mapping.values()}
     skipped_variant_types = []
+
+    included_kbvariant_ids = list(set([item['kbVariantId'] for item in included_kb_matches]))
+
     for kb_match in kb_matches:
+        if kb_match['kbVariantId'] not in included_kbvariant_ids:
+            continue
         variant_type = kb_match["variantType"]
         variant_key = kb_match["variant"]
         if kb_match["category"] == "unknown":

pori_python/ipr/main.py

@@ -512,8 +512,9 @@ def ipr_report(
     )
 
     # KEY ALTERATIONS
-    # must do after pruning of kbMatches for kb_matched_sections
-    key_alterations, variant_counts = create_key_alterations(gkb_matches, all_variants)
+    key_alterations, variant_counts = create_key_alterations(
+        gkb_matches, all_variants, kb_matched_sections['kbMatches']
+    )
 
     # OUTPUT CONTENT
     # thread safe deep-copy the original content

tests/test_ipr/test_ipr.py

@@ -11,6 +11,7 @@
     get_kb_statement_matched_conditions,
     get_kb_variants,
     get_kb_matches_sections,
+    create_key_alterations,
 )
 from pori_python.types import Statement
 
@@ -484,9 +485,9 @@ def test_germline_kb_matches(self):
         "kbContextId": "#135:8764",
         "kbRelevanceId": "#147:32",
         "kbStatementId": "#155:13511",
-        "requiredKbMatches": ["#159:5426", "#161:938"],
+        "requiredKbMatches": ["#159:54261", "#161:9381"],
         "kbVariant": "BRCA1 mutation",
-        "kbVariantId": "#161:938",
+        "kbVariantId": "#161:9381",
         "matchedCancer": False,
         "reference": "MOAlmanac FDA-56",
         "relevance": "therapy",
@@ -495,6 +496,13 @@ def test_germline_kb_matches(self):
     },
 ]
 
+ALL_VARIANTS = [
+    {"variant": "var1", "key": '1', "variantType": 'mut'},
+    {"variant": "var2", "key": '2', "variantType": 'mut'},
+    {"variant": "var3", "key": '3', "variantType": 'mut'},
+    {"variant": "var4", "key": '4', "variantType": 'mut'},
+]
+
 BASIC_GKB_MATCH = {
     "approvedTherapy": False,
     "category": "test",
@@ -830,3 +838,21 @@ def test_partial_matches_included(self):
         kbcs = get_kb_statement_matched_conditions(gkb_matches, allow_partial_matches=True)
         assert len(stmts) == 2  # X and Y
         assert len(kbcs) == 2
+
+    def test_create_key_alterations_includes_only_pruned_kbmatches(self):
+        gkb_matches = create_gkb_matches(GKB_MATCHES)
+
+        sections1 = get_kb_matches_sections(gkb_matches, allow_partial_matches=False)
+        key_alts1, counts1 = create_key_alterations(
+            gkb_matches, ALL_VARIANTS, sections1['kbMatches']
+        )
+
+        sections2 = get_kb_matches_sections(gkb_matches, allow_partial_matches=True)
+        key_alts2, counts2 = create_key_alterations(
+            gkb_matches, ALL_VARIANTS, sections2['kbMatches']
+        )
+
+        # check partial-match-only variants are not included in key alterations when
+        # partial matches is false
+        assert len(key_alts1) == 3
+        assert len(key_alts2) == 4