Merge branch 'feat/DEVSU-2797-remove-gsc-defaults_dustin' of https://github.com/bcgsc/pori_python into improvement/SDEV-5193_mypy_type_fixes

Author: dustinbleile

.github/workflows/pytest.yml

@@ -46,6 +46,7 @@ jobs:
         IPR_PASS: ${{ secrets.IPR_TEST_PASSWORD }}
         GRAPHKB_USER: ${{ secrets.GKB_TEST_USER }}
         GRAPHKB_PASS: ${{ secrets.GKB_TEST_PASS }}
+        GRAPHKB_URL: ${{ secrets.GKB_TEST_URL }}
         # SDEV-3381 - Turn off integration tests temporarily, till efficiency is increased
         # turn on integration tests for one python version only
         EXCLUDE_INTEGRATION_TESTS: ${{ matrix.python-version != '3.11' }}

.github/workflows/quick-pytest.yml

@@ -39,8 +39,10 @@ jobs:
       env:
         IPR_USER: ${{ secrets.IPR_TEST_USER }}
         IPR_PASS: ${{ secrets.IPR_TEST_PASSWORD }}
+        IPR_URL: ${{ secrets.IPR_TEST_URL }}
         GRAPHKB_USER: ${{ secrets.GKB_TEST_USER }}
         GRAPHKB_PASS: ${{ secrets.GKB_TEST_PASS }}
+        GRAPHKB_URL: ${{ secrets.GKB_TEST_URL }}
         EXCLUDE_INTEGRATION_TESTS: 1
         # EXCLUDE_INTEGRATION_TESTS: ${{ matrix.python-version != '3.11' }}
-      if: github.event_name != 'pull_request'
+      if: github.event_name != 'pull_request'

pori_python/graphkb/__init__.py

@@ -1,2 +1 @@
-from .constants import DEFAULT_URL  # noqa: F401
 from .util import GraphKBConnection, logger  # noqa: F401

pori_python/graphkb/constants.py

@@ -4,12 +4,7 @@
 from pori_python.types import CategoryBaseTermMapping
 
 DEFAULT_LIMIT = 1000
-GKB_BASE_URL = "https://graphkb-api.bcgsc.ca/api"
-GKB_STAGING_URL = "https://graphkbstaging-api.bcgsc.ca/api"
-GKB_DEV_URL = "https://graphkbdev-api.bcgsc.ca/api"
-DEFAULT_URL = GKB_BASE_URL
 
-PREFERRED_GENE_SOURCE = "#39:5"  # HGNC
 PREFERRED_GENE_SOURCE_NAME = "HGNC"
 
 BASE_RETURN_PROPERTIES = ["@rid", "@class"]

pori_python/graphkb/genes.py

@@ -17,7 +17,6 @@
     GSC_PHARMACOGENOMIC_SOURCE_EXCLUDE_LIST,
     ONCOGENE,
     ONCOKB_SOURCE_NAME,
-    PREFERRED_GENE_SOURCE,
     PREFERRED_GENE_SOURCE_NAME,
     RELEVANCE_BASE_TERMS,
     TSO500_SOURCE_NAME,
@@ -267,24 +266,23 @@ def get_cancer_predisposition_info(
 
 
 def get_gene_linked_cancer_predisposition_info(
-    conn: GraphKBConnection, source: str = PREFERRED_GENE_SOURCE
+    conn: GraphKBConnection, source: str = PREFERRED_GENE_SOURCE_NAME
 ) -> Tuple[List[str], Dict[str, Tuple[str, List[str]]]]:
     """
     Return two lists from GraphKB, one of cancer predisposition genes and one of associated variants.
 
     GERO-272 - criteria for what counts as a "cancer predisposition" variant
 
     In short:
-    * Statement 'source' is 'CGL'
+    * Statement 'source' is 'CGL' (not related to the preferred gene source)
     * Statement 'relevance' is 'pathogenic'
     * gene is gotten from any associated 'PositionalVariant' records
 
     Example: https://graphkb.bcgsc.ca/view/Statement/155:11616
 
-
-
     Returns:
         genes: list of cancer predisposition genes
+            (using names from the source specified in this function's arguments)
         variants: dictionary mapping pharmacogenomic variant IDs to variant display names
     """
     genes = set()
@@ -372,7 +370,7 @@ def get_pharmacogenomic_info(
 
 
 def get_gene_linked_pharmacogenomic_info(
-    conn: GraphKBConnection, source: str = PREFERRED_GENE_SOURCE
+    conn: GraphKBConnection, source: str = PREFERRED_GENE_SOURCE_NAME
 ) -> Tuple[List[str], Dict[str, Tuple[str, List[str]]]]:
     """
     Return two lists from GraphKB, one of pharmacogenomic genes and one of associated variants.

pori_python/graphkb/util.py

@@ -14,7 +14,7 @@
 
 from pori_python.types import ParsedVariant, PositionalVariant, Record
 
-from .constants import DEFAULT_LIMIT, DEFAULT_URL, TYPES_TO_NOTATION, AA_3to1_MAPPING
+from .constants import DEFAULT_LIMIT, TYPES_TO_NOTATION, AA_3to1_MAPPING
 
 QUERY_CACHE: Dict[Any, Any] = {}
 
@@ -98,7 +98,7 @@ def cache_key(request_body) -> str:
 class GraphKBConnection:
     def __init__(
         self,
-        url: str = os.environ.get("GRAPHKB_URL", DEFAULT_URL),
+        url: str = os.environ.get("GRAPHKB_URL", ""),
         username: str = "",
         password: str = "",
         use_global_cache: bool = True,
@@ -143,6 +143,8 @@ def request(self, endpoint: str, method: str = "GET", **kwargs) -> Dict:
         Returns:
             dict: the json response as a python dict
         """
+        if not self.url:
+            raise ValueError("no GraphKBConnection url set - cannot make a login demo")
         url = join_url(self.url, endpoint)
         self.request_count += 1
         connect_timeout = 7
@@ -222,6 +224,8 @@ def login_demo(self) -> None:
         1. get a first token from KeyCloak using username and password; self.login_demo()
         2. get a second token from the GraphKB API using keyCloakToken; self.login()
         """
+        if not self.url:
+            raise ValueError("no GraphKBConnection url set - cannot make a login demo")
         url_parts = urlsplit(self.url)
         base_url = f"{url_parts.scheme}://{url_parts.netloc}"
 
@@ -250,8 +254,11 @@ def login(self, username: str, password: str, pori_demo: bool = False) -> None:
         connect_timeout = 7
         read_timeout = 61
 
-        # KBDEV-1328. Alt. GraphKB login for GSC's PORI online demo
-        if pori_demo or "pori-demo" in self.url:
+        if not self.url:
+            raise ValueError("no GraphKBConnection url set - cannot login")
+        elif pori_demo or "pori-demo" in self.url:
+            # KBDEV-1328. Alt. GraphKB login for GSC's PORI online demo
+            logger.warning("login demo")
             self.login_demo()
 
         # use requests package directly to avoid recursion loop on login failure

pori_python/ipr/connection.py

@@ -6,7 +6,6 @@
 import zlib
 from typing import Dict, List
 
-from .constants import DEFAULT_URL
 from .util import logger
 
 IMAGE_MAX = 20  # cannot upload more than 20 images at a time
@@ -17,7 +16,7 @@ def __init__(
         self,
         username: str,
         password: str,
-        url: str = os.environ.get("IPR_URL", DEFAULT_URL),
+        url: str = os.environ.get("IPR_URL"),
     ):
         self.token = None
         self.url = url

pori_python/ipr/constants.py

@@ -1,4 +1,3 @@
-DEFAULT_URL = "https://iprstaging-api.bcgsc.ca/api"
 GERMLINE_BASE_TERMS = ("pharmacogenomic", "cancer predisposition")  # based on graphkb.constants
 VARIANT_CLASSES = {"Variant", "CategoryVariant", "PositionalVariant", "CatalogueVariant"}
 
@@ -26,3 +25,16 @@
         "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

@@ -25,8 +25,8 @@
 
 from .constants import (
     COSMIC_SIGNATURE_VARIANT_TYPE,
-    DEFAULT_URL,
     HLA_SIGNATURE_VARIANT_TYPE,
+    HRD_MAPPING,
     MSI_MAPPING,
     TMB_SIGNATURE,
     TMB_SIGNATURE_VARIANT_TYPE,
@@ -38,7 +38,7 @@
 SPECIFICATION = os.path.join(os.path.dirname(__file__), "content.spec.json")
 
 # content in the local specification should match the values in IPR_API_SPEC_JSON_URL
-IPR_API_SPEC_JSON_URL = f'{os.environ.get("IPR_URL", DEFAULT_URL)}/spec.json'
+IPR_API_SPEC_JSON_URL = f'{os.environ.get("IPR_URL")}/spec.json'
 
 # TODO: GERO-307 - use SPECIFICATION json to derive the variant required and optional details defined below
 
@@ -560,6 +560,23 @@ def preprocess_msi(msi: Any) -> Sequence[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],