fix: handle brand name lookup KeyError (#404)

src/therapy/etl/hemonc.py

@@ -129,7 +129,15 @@ def _get_rels(self, therapies: Dict, brand_names: Dict, conditions: Dict) -> Dic
                     _logger.warning(f"Unrecognized `Maps To` source: {src_raw}")
 
             elif rel_type == "Has brand name":
-                record["trade_names"].append(brand_names[row[1]])
+                try:
+                    record["trade_names"].append(brand_names[row[1]])
+                except KeyError:
+                    _logger.warning(
+                        "Unrecognized brand name ID (%s) for HemOnc concept %s",
+                        row[1],
+                        row[0],
+                    )
+                    continue
 
             elif rel_type == "Was FDA approved yr":
                 try:

tests/data/fixtures/merged_fixtures.json

@@ -185,6 +185,7 @@
     ],
     "has_indication": [
       "[\"hemonc:671\", \"Testicular cancer\", \"ncit:C7251\", {\"regulatory_body\": \"FDA\"}]",
+      "[\"hemonc:645\", \"Ovarian cancer\", \"ncit:C7431\", {\"regulatory_body\": \"FDA\"}]",
       "[\"hemonc:569\", \"Bladder cancer\", \"ncit:C9334\", {\"regulatory_body\": \"FDA\"}]",
       "[\"mesh:D018273\", \"Carcinoma, Islet Cell\", \"ncit:C3770\", {\"chembl_max_phase_for_ind\": \"chembl_phase_2\"}]",
       "[\"mesh:D000077195\", \"Squamous Cell Carcinoma of Head and Neck\", \"ncit:C4044\", {\"chembl_max_phase_for_ind\": \"chembl_phase_3\"}]",

tests/data/fixtures/query_fixtures.json

@@ -75,6 +75,27 @@
               ],
               "type": "Disease"
             },
+            {
+              "id": "hemonc:645",
+              "label": "Ovarian cancer",
+              "extensions": [
+                {
+                  "type": "Extension",
+                  "name": "regulatory_body",
+                  "value": "FDA"
+                }
+              ],
+              "mappings": [
+                {
+                  "coding": {
+                    "system": "ncit",
+                    "code": "C7431"
+                  },
+                  "relation": "relatedMatch"
+                }
+              ],
+              "type": "Disease"
+            },
             {
               "id": "hemonc:569",
               "label": "Bladder cancer",
@@ -3500,7 +3521,7 @@
       "TX",
       "TREAT",
       "Therapeutic Method",
-            "Therapeutic",
+      "Therapeutic",
       "Treatment",
       "Therapeutic Technique",
       "Therapy",
@@ -3532,7 +3553,7 @@
               "treatment",
               "TREAT",
               "Therapy",
-            "Therapeutic",
+              "Therapeutic",
               "Therapeutic Method",
               "Therapeutic Interventions",
               "Treatment",
@@ -3851,6 +3872,12 @@
                 "supplemental_info": {
                   "regulatory_body": "FDA"
                 }
+              },
+              {
+                "disease_id": "hemonc:645",
+                "disease_label": "Ovarian cancer",
+                "normalized_disease_id": "ncit:C7431",
+                "supplemental_info": { "regulatory_body": "FDA" }
               }
             ]
           }

tests/data/hemonc/hemonc_concepts_2023-09-07.csv → tests/data/hemonc/hemonc_concepts_2023-12-17.csv

@@ -11,6 +11,7 @@ Testicular cancer,HemOnc,Condition,671,2019-05-27,2099-12-31,
 1978,HemOnc,Year,738,2019-05-27,2099-12-31,
 1995,HemOnc,Year,755,2019-05-27,2099-12-31,
 2008,HemOnc,Year,768,2019-05-27,2099-12-31,
+2018,HemOnc,Year,778,2019-05-27,2099-12-31,
 Cytofos,HemOnc,Brand Name,3100,2019-05-27,2099-12-31,
 Ethyol,HemOnc,Brand Name,3101,2019-05-27,2099-12-31,
 Bendamax,HemOnc,Brand Name,5159,2019-05-27,2099-12-31,
@@ -91,6 +92,7 @@ Tecnoplatin,HemOnc,Brand Name,13531,2019-05-27,2099-12-31,
 Tisplal,HemOnc,Brand Name,13532,2019-05-27,2099-12-31,
 Unistin,HemOnc,Brand Name,13533,2019-05-27,2099-12-31,
 Firmagon,HemOnc,Brand Name,15844,2019-05-27,2099-12-31,
+Nivestim,HemOnc,Brand Name,18918,2019-05-27,2021-09-06,D
 Intravenous,HemOnc,Route,44957,2019-05-27,2099-12-31,
 Irritant,HemOnc,Component Class,44997,2019-05-27,2099-12-31,
 WHO Essential Cancer Medicine,HemOnc,Component Class,45050,2019-05-27,2099-12-31,
@@ -99,14 +101,18 @@ Chemotherapy protective agent,HemOnc,Component Class,45069,2019-05-27,2099-12-31
 Vesicant,HemOnc,Component Class,45092,2019-05-27,2099-12-31,
 Subcutaneous,HemOnc,Route,45153,2019-05-27,2099-12-31,
 Nitrogen mustard,HemOnc,Component Class,45230,2019-05-27,2099-12-31,
+Biosimilar,HemOnc,Component Class,45268,2019-05-27,2099-12-31,
 Intracavitary,HemOnc,Route,45273,2019-05-27,2099-12-31,
 Light-chain (AL) amyloidosis medication,HemOnc,Component Class,45303,2019-05-27,2099-12-31,
 Platinum agent,HemOnc,Component Class,45388,2019-05-27,2099-12-31,
 Citosin,HemOnc,Brand Name,45480,2019-07-22,2099-12-31,
 GnRH antagonist,HemOnc,Component Class,45632,2019-05-27,2099-12-31,
+Granulocyte colony-stimulating factor,HemOnc,Component Class,45734,2019-05-27,2099-12-31,
 Indolent lymphoma,HemOnc,Condition,46094,2019-05-27,2099-12-31,
 Androgen receptor inhibitor,HemOnc,Component Class,46196,2019-05-27,2099-12-31,
 Belrapzo,HemOnc,Brand Name,51576,2022-06-26,2099-12-31,
+Filgrastim-aafi,HemOnc,Component,66258,2021-12-08,2099-12-31,
+Nivestym,HemOnc,Brand Name,69510,2021-12-08,2099-12-31,
 Cysplatyna,HemOnc,Brand Name,76972,2019-07-22,2099-12-31,
 Peyrone's Chloride,HemOnc,Brand Name,93467,2019-07-22,2099-12-31,
 Peyrone's Salt,HemOnc,Brand Name,93468,2019-07-22,2099-12-31,

tests/data/hemonc/hemonc_rels_2023-09-07.csv → tests/data/hemonc/hemonc_rels_2023-12-17.csv

@@ -1,14 +1,14 @@
 concept_code_1,concept_code_2,vocabulary_id_1,vocabulary_id_2,relationship_id
 26,3100,HemOnc,HemOnc,Has brand name
 26,3101,HemOnc,HemOnc,Has brand name
+26,645,HemOnc,HemOnc,Has FDA indication
 26,44957,HemOnc,HemOnc,Has route
 26,45069,HemOnc,HemOnc,Is a
 26,755,HemOnc,HemOnc,Was FDA approved yr
-26,4126,HemOnc,RxNorm,Maps to
-26,J0207,HemOnc,HCPCS,Maps to
 26,45069,HemOnc,HemOnc,Has major class
+26,J0207,HemOnc,HCPCS,Maps to
 26,76310-017,HemOnc,NDC,Maps to
-26,645,HemOnc,HemOnc,Has FDA indication
+26,4126,HemOnc,RxNorm,Maps to
 65,51576,HemOnc,HemOnc,Has brand name
 65,5159,HemOnc,HemOnc,Has brand name
 65,5160,HemOnc,HemOnc,Has brand name
@@ -24,30 +24,30 @@ concept_code_1,concept_code_2,vocabulary_id_1,vocabulary_id_2,relationship_id
 65,5170,HemOnc,HemOnc,Has brand name
 65,5171,HemOnc,HemOnc,Has brand name
 65,5172,HemOnc,HemOnc,Has brand name
+65,581,HemOnc,HemOnc,Has FDA indication
+65,46094,HemOnc,HemOnc,Has FDA indication
 65,44957,HemOnc,HemOnc,Has route
 65,45092,HemOnc,HemOnc,Is a
 65,45230,HemOnc,HemOnc,Is a
 65,45058,HemOnc,HemOnc,Is a
 65,45303,HemOnc,HemOnc,Is a
 65,45050,HemOnc,HemOnc,Is a
 65,768,HemOnc,HemOnc,Was FDA approved yr
-65,134547,HemOnc,RxNorm,Maps to
+65,45058,HemOnc,HemOnc,Has major class
+65,45230,HemOnc,HemOnc,Has minor class
 65,C9042,HemOnc,HCPCS,Maps to
 65,C9243,HemOnc,HCPCS,Maps to
 65,J9033,HemOnc,HCPCS,Maps to
 65,J9034,HemOnc,HCPCS,Maps to
-65,J9036,HemOnc,HCPCS,Maps to
 65,J9056,HemOnc,HCPCS,Maps to
 65,J9058,HemOnc,HCPCS,Maps to
 65,J9059,HemOnc,HCPCS,Maps to
-65,45058,HemOnc,HemOnc,Has major class
-65,45230,HemOnc,HemOnc,Has minor class
+65,J9036,HemOnc,HCPCS,Maps to
 65,42367-521,HemOnc,NDC,Maps to
 65,63459-348,HemOnc,NDC,Maps to
 65,63459-390,HemOnc,NDC,Maps to
 65,63459-391,HemOnc,NDC,Maps to
-65,581,HemOnc,HemOnc,Has FDA indication
-65,46094,HemOnc,HemOnc,Has FDA indication
+65,134547,HemOnc,RxNorm,Maps to
 105,13471,HemOnc,HemOnc,Has brand name
 105,13472,HemOnc,HemOnc,Has brand name
 105,13473,HemOnc,HemOnc,Has brand name
@@ -117,6 +117,9 @@ concept_code_1,concept_code_2,vocabulary_id_1,vocabulary_id_2,relationship_id
 105,13531,HemOnc,HemOnc,Has brand name
 105,13532,HemOnc,HemOnc,Has brand name
 105,13533,HemOnc,HemOnc,Has brand name
+105,569,HemOnc,HemOnc,Has FDA indication
+105,645,HemOnc,HemOnc,Has FDA indication
+105,671,HemOnc,HemOnc,Has FDA indication
 105,44957,HemOnc,HemOnc,May have route
 105,45273,HemOnc,HemOnc,May have route
 105,45092,HemOnc,HemOnc,Is a
@@ -125,12 +128,11 @@ concept_code_1,concept_code_2,vocabulary_id_1,vocabulary_id_2,relationship_id
 105,45058,HemOnc,HemOnc,Is a
 105,45050,HemOnc,HemOnc,Is a
 105,738,HemOnc,HemOnc,Was FDA approved yr
-105,2555,HemOnc,RxNorm,Maps to
+105,45058,HemOnc,HemOnc,Has major class
+105,45388,HemOnc,HemOnc,Has minor class
 105,C9418,HemOnc,HCPCS,Maps to
 105,J9060,HemOnc,HCPCS,Maps to
 105,J9062,HemOnc,HCPCS,Maps to
-105,45058,HemOnc,HemOnc,Has major class
-105,45388,HemOnc,HemOnc,Has minor class
 105,0143-9504,HemOnc,NDC,Maps to
 105,0143-9505,HemOnc,NDC,Maps to
 105,0703-5747,HemOnc,NDC,Maps to
@@ -145,17 +147,29 @@ concept_code_1,concept_code_2,vocabulary_id_1,vocabulary_id_2,relationship_id
 105,68083-162,HemOnc,NDC,Maps to
 105,68083-163,HemOnc,NDC,Maps to
 105,70860-206,HemOnc,NDC,Maps to
-105,569,HemOnc,HemOnc,Has FDA indication
-105,671,HemOnc,HemOnc,Has FDA indication
+105,2555,HemOnc,RxNorm,Maps to
 151,15844,HemOnc,HemOnc,Has brand name
 151,120237,HemOnc,HemOnc,Has brand name
+151,658,HemOnc,HemOnc,Has FDA indication
 151,45153,HemOnc,HemOnc,Has route
 151,45632,HemOnc,HemOnc,Is a
 151,768,HemOnc,HemOnc,Was FDA approved yr
-151,475230,HemOnc,RxNorm,Maps to
-151,J9155,HemOnc,HCPCS,Maps to
 151,46196,HemOnc,HemOnc,Has major class
 151,45632,HemOnc,HemOnc,Has minor class
+151,J9155,HemOnc,HCPCS,Maps to
 151,55566-8303,HemOnc,NDC,Maps to
 151,55566-8403,HemOnc,NDC,Maps to
-151,658,HemOnc,HemOnc,Has FDA indication
+151,475230,HemOnc,RxNorm,Maps to
+66258,18918,HemOnc,HemOnc,Has brand name
+66258,69510,HemOnc,HemOnc,Has brand name
+66258,45153,HemOnc,HemOnc,Has route
+66258,45734,HemOnc,HemOnc,Is a
+66258,45268,HemOnc,HemOnc,Is a
+66258,45050,HemOnc,HemOnc,Is a
+66258,778,HemOnc,HemOnc,Was FDA approved yr
+66258,Q5110,HemOnc,HCPCS,Maps to
+66258,0069-0291,HemOnc,NDC,Maps to
+66258,0069-0292,HemOnc,NDC,Maps to
+66258,0069-0293,HemOnc,NDC,Maps to
+66258,0069-0294,HemOnc,NDC,Maps to
+66258,68442,HemOnc,RxNorm,Maps to

tests/data/hemonc/hemonc_synonyms_2023-09-07.csv → tests/data/hemonc/hemonc_synonyms_2023-12-17.csv

@@ -29,3 +29,4 @@ ASP-3550,151,HemOnc
 FE-200486,151,HemOnc
 ASP 3550,151,HemOnc
 FE 200486,151,HemOnc
+Filgrastim-aafi,66258,HemOnc

tests/scripts/build_hemonc_data.py

@@ -5,7 +5,7 @@
 from therapy.database import create_db
 from therapy.etl import HemOnc
 
-TEST_IDS = ["65", "105", "151", "26"]
+TEST_IDS = ["65", "105", "151", "26", "66258"]
 
 ho = HemOnc(create_db())  # don't need to write any data
 ho._extract_data(False)

tests/unit/test_hemonc.py

@@ -50,6 +50,12 @@ def cisplatin():
                     "normalized_disease_id": "ncit:C7251",
                     "supplemental_info": {"regulatory_body": "FDA"},
                 },
+                {
+                    "disease_id": "hemonc:645",
+                    "disease_label": "Ovarian cancer",
+                    "normalized_disease_id": "ncit:C7431",
+                    "supplemental_info": {"regulatory_body": "FDA"},
+                },
             ],
         }
     )
@@ -146,7 +152,25 @@ def degarelix():
     )
 
 
-def test_concept_id_match(hemonc, compare_response, cisplatin, bendamustine, degarelix):
+@pytest.fixture(scope="module")
+def filgrastim():
+    """Create fixture for filgrastim drug (tests handling of deprecated brand name)"""
+    return Therapy(
+        **{
+            "label": "Filgrastim-aafi",
+            "concept_id": "hemonc:66258",
+            "aliases": [],
+            "xrefs": ["rxcui:68442"],
+            "trade_names": ["Nivestym"],
+            "approval_ratings": ["hemonc_approved"],
+            "approval_year": ["2018"],
+        }
+    )
+
+
+def test_concept_id_match(
+    hemonc, compare_response, cisplatin, bendamustine, degarelix, filgrastim
+):
     """Test that concept ID queries resolve to correct record."""
     response = hemonc.search("hemonc:105")
     compare_response(response, MatchType.CONCEPT_ID, cisplatin)
@@ -157,8 +181,13 @@ def test_concept_id_match(hemonc, compare_response, cisplatin, bendamustine, deg
     response = hemonc.search("hemonc:151")
     compare_response(response, MatchType.CONCEPT_ID, degarelix)
 
+    response = hemonc.search("hemonc:66258")
+    compare_response(response, MatchType.CONCEPT_ID, filgrastim)
 
-def test_label_match(hemonc, compare_response, cisplatin, bendamustine, degarelix):
+
+def test_label_match(
+    hemonc, compare_response, cisplatin, bendamustine, degarelix, filgrastim
+):
     """Test that label queries resolve to correct record."""
     response = hemonc.search("cisplatin")
     compare_response(response, MatchType.LABEL, cisplatin)
@@ -169,6 +198,9 @@ def test_label_match(hemonc, compare_response, cisplatin, bendamustine, degareli
     response = hemonc.search("DEGARELIX")
     compare_response(response, MatchType.LABEL, degarelix)
 
+    response = hemonc.search("Filgrastim-aafi")
+    compare_response(response, MatchType.LABEL, filgrastim)
+
 
 def test_alias_match(hemonc, compare_response, cisplatin, bendamustine, degarelix):
     """Test that alias queries resolve to correct record."""
@@ -207,7 +239,9 @@ def test_trade_name(hemonc, compare_response, bendamustine, degarelix):
     assert response.match_type == MatchType.NO_MATCH
 
 
-def test_xref_match(hemonc, compare_response, cisplatin, bendamustine, degarelix):
+def test_xref_match(
+    hemonc, compare_response, cisplatin, bendamustine, degarelix, filgrastim
+):
     """Test that xref query resolves to correct record."""
     response = hemonc.search("rxcui:2555")
     compare_response(response, MatchType.XREF, cisplatin)
@@ -218,6 +252,9 @@ def test_xref_match(hemonc, compare_response, cisplatin, bendamustine, degarelix
     response = hemonc.search("rxcui:475230")
     compare_response(response, MatchType.XREF, degarelix)
 
+    response = hemonc.search("rxcui:68442")
+    compare_response(response, MatchType.XREF, filgrastim)
+
 
 def test_metadata(hemonc):
     """Test that source metadata returns correctly."""