Merge branch 'develop' into feat/DEVSU-2786-get-trial-nct-ids

Author: elewis2

pori_python/ipr/constants.py

@@ -26,3 +26,16 @@
         "variantTypeName": "signature present",
     },
 }
+# Mapping hrd from pipeline terms to GraphKB terms
+HRD_MAPPING = {
+    "homologous recombination deficiency strong signature": {
+        "displayName": "homologous recombination deficiency strong signature",
+        "signatureName": "homologous recombination deficiency",
+        "variantTypeName": "strong signature",
+    },
+    "homologous recombination deficiency moderate signature": {
+        "displayName": "homologous recombination deficiency moderate signature",
+        "signatureName": "homologous recombination deficiency",
+        "variantTypeName": "moderate signature",
+    },
+}

pori_python/ipr/content.spec.json

@@ -541,6 +541,20 @@
             },
             "type": "array"
         },
+        "hrd": {
+            "properties": {
+                "kbCategory": {
+                    "type": "string"
+                },
+                "score": {
+                    "type": "number"
+                }
+            },
+            "required": [
+                "score"
+            ],
+            "type": "object"
+        },
         "images": {
             "items": {
                 "example": {

pori_python/ipr/inputs.py

@@ -28,6 +28,7 @@
     DEFAULT_URL,
     HLA_SIGNATURE_VARIANT_TYPE,
     MSI_MAPPING,
+    HRD_MAPPING,
     TMB_SIGNATURE,
     TMB_SIGNATURE_VARIANT_TYPE,
 )
@@ -557,6 +558,23 @@ def preprocess_msi(msi: Any) -> Iterable[Dict]:
     return []
 
 
+def preprocess_hrd(hrd: Any) -> Iterable[Dict]:
+    """
+    Process hrd input into preformatted signature input.
+    HRD gets mapped to corresponding GraphKB Signature CategoryVariants.
+    """
+    if hrd:
+        hrd_cat = hrd.get("kbCategory", "")
+
+        hrd_variant = HRD_MAPPING.get(hrd_cat, None)
+
+        # Signature CategoryVariant created either for msi or mss
+        if hrd_variant:
+            return [hrd_variant]
+
+    return []
+
+
 def check_variant_links(
     small_mutations: List[IprSmallMutationVariant],
     expression_variants: List[IprExprVariant],

pori_python/ipr/main.py

@@ -32,6 +32,7 @@
     preprocess_expression_variants,
     preprocess_hla,
     preprocess_msi,
+    preprocess_hrd,
     preprocess_signature_variants,
     preprocess_small_mutations,
     preprocess_structural_variants,
@@ -387,6 +388,7 @@ def ipr_report(
                 content.get("genomeTmb", ""),  # newer tmb pipeline
             ),
             *preprocess_msi(content.get("msi", None)),
+            *preprocess_hrd(content.get("hrd", None)),
         ]
     )
     small_mutations: List[IprSmallMutationVariant] = preprocess_small_mutations(
@@ -515,7 +517,6 @@ def ipr_report(
     # OUTPUT CONTENT
     # thread safe deep-copy the original content
     output = json.loads(json.dumps(content))
-
     output.update(kb_matched_sections)
     output.update(
         {
@@ -547,6 +548,13 @@ def ipr_report(
     )
     output.setdefault("images", []).extend(select_expression_plots(gkb_matches, all_variants))
 
+    # if input includes hrdScore field, that is ok to pass to db
+    # but prefer the 'hrd' field if it exists
+    if output.get('hrd'):
+        if output.get('hrd').get('score'):
+            output['hrdScore'] = output['hrd']['score']
+        output.pop('hrd')  # kbmatches have already been made
+
     ipr_spec = ipr_conn.get_spec()
     output = clean_unsupported_content(output, ipr_spec)
     ipr_result = {}

tests/test_ipr/test_inputs.py

@@ -8,6 +8,7 @@
 from pori_python.graphkb.match import INPUT_COPY_CATEGORIES
 from pori_python.ipr.constants import (
     MSI_MAPPING,
+    HRD_MAPPING,
     TMB_SIGNATURE,
     TMB_SIGNATURE_HIGH_THRESHOLD,
 )
@@ -21,6 +22,7 @@
     preprocess_expression_variants,
     preprocess_hla,
     preprocess_msi,
+    preprocess_hrd,
     preprocess_signature_variants,
     preprocess_small_mutations,
     preprocess_structural_variants,
@@ -49,6 +51,9 @@
 }
 EXPECTED_TMB = {TMB_SIGNATURE}
 EXPECTED_MSI = {MSI_MAPPING.get("microsatellite instability")["signatureName"]}
+EXPECTED_HRD = {
+    HRD_MAPPING.get("homologous recombination deficiency strong signature")["signatureName"]
+}
 
 
 def read_data_file(filename):
@@ -234,6 +239,13 @@ class TestPreProcessSignatureVariants:
             }
         ]
     )
+    hrd = preprocess_hrd(
+        {
+            "score": 9999,
+            "kbCategory": "homologous recombination deficiency strong signature",
+            "key": "homologous recombination deficiency strong signature",
+        }
+    )
 
     # tests on preprocessed records
     def test_preprocess_cosmic(self) -> None:
@@ -272,6 +284,15 @@ def test_preprocess_msi(self) -> None:
         signatureNames = {r.get("signatureName", "") for r in self.msi}
         assert len(EXPECTED_MSI.symmetric_difference(signatureNames)) == 0
 
+    def test_preprocess_hrd(self) -> None:
+        assert self.hrd
+        assert len(self.hrd) == len(EXPECTED_HRD)
+        assert "variantTypeName" in self.hrd[0]
+        assert "displayName" in self.hrd[0]
+
+        signatureNames = {r.get("signatureName", "") for r in self.hrd}
+        assert len(EXPECTED_HRD.symmetric_difference(signatureNames)) == 0
+
     def test_preprocess_signature_variants(self) -> None:
         records = preprocess_signature_variants(
             [

tests/test_ipr/test_main.py

@@ -66,6 +66,16 @@ def report_upload_content(tmp_path_factory) -> Dict:
                     pd.read_csv(get_test_file("fusions.tab"), sep="\t").to_json(orient="records")
                 ),
                 "kbDiseaseMatch": "colorectal cancer",
+                "msi": [
+                    {
+                        "score": 1000.0,
+                        "kbCategory": "microsatellite instability",
+                    }
+                ],
+                "hrd": {
+                    "score": 9999.0,
+                    "kbCategory": "homologous recombination deficiency strong signature",
+                },
             },
             allow_nan=False,
         )

tests/test_ipr/test_upload.py

@@ -66,6 +66,16 @@ def loaded_reports(tmp_path_factory) -> Generator:
                 "collectionDate": "12-12-12",
             },
         ],
+        "msi": [
+            {
+                "score": 1000.0,
+                "kbCategory": "microsatellite instability",
+            }
+        ],
+        "hrd": {
+            "score": 9999.0,
+            "kbCategory": "homologous recombination deficiency strong signature",
+        },
         "expressionVariants": json.loads(
             pd.read_csv(get_test_file("expression.short.tab"), sep="\t").to_json(orient="records")
         ),