phc-ingestion 0.10.12__tar.gz → 0.10.14__tar.gz

{phc-ingestion-0.10.12 → phc-ingestion-0.10.14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phc-ingestion
-Version: 0.10.12
+Version: 0.10.14
 Summary: Functions for LifeOmic PHC genomic ingestions
 License: MIT
 Author-email: LifeOmic Development <development@lifeomic.com>

{phc-ingestion-0.10.12 → phc-ingestion-0.10.14}/ingestion/nebula/manifest_assembler.py

@@ -1,6 +1,10 @@
-from typing import Optional, TypedDict, Any
-from datetime import datetime
+import re
+import gzip
+from typing import Optional, TypedDict, Any, Callable
+from datetime import datetime, timezone
+from dateutil import parser
 from ingestion.shared_util.lambda_client import LambdaClient
+from ingestion.vcf_standardization.util.read_write import read_headers
 from ingestion.nebula.constants import (
     DATASET_SYSTEM,
     NEBULA_KIT_ID_SYSTEM,
@@ -9,7 +13,6 @@ from ingestion.nebula.constants import (
     NEBULA_BODY_SITE,
     NEBULA_INDICATION,
 )
-import re
 
 
 class HumanName(TypedDict):
@@ -49,11 +52,14 @@ class Practitioner(Resource):
 
 
 class ManifestAssembler:
-    def __init__(self, ingestion_id: str, account_id: str, project_id: str, kit_id: str):
+    def __init__(
+        self, ingestion_id: str, account_id: str, project_id: str, kit_id: str, vcf_file_path: str
+    ):
         self.ingestion_id = ingestion_id
         self.account_id = account_id
         self.project_id = project_id
         self.kit_id = kit_id
+        self.vcf_file_path = vcf_file_path
         self.client = LambdaClient(
             "patient-service",
             {
@@ -142,6 +148,37 @@ class ManifestAssembler:
             "fullName": f"{first_name} {last_name}",
         }
 
+    def _safe(self, lambda_func: Callable[[], Any]) -> Any:
+        try:
+            return lambda_func()
+        except Exception as e:
+            return None
+
+    def __extract_collection_date(self) -> dict[str, str]:
+        in_file = self.vcf_file_path
+        with gzip.open(in_file, "rt") if in_file.endswith(".gz") else open(in_file, "r") as f:
+            headers = read_headers(f)
+            for header in headers:
+                epoch_parts = re.search(r"Epoch=(\d+)", header)
+                epoch_time_raw = epoch_parts.group(1) if epoch_parts else None
+                epoch_time = self._safe(
+                    lambda: datetime.fromtimestamp(int(epoch_time_raw) / 1000.0, tz=timezone.utc)
+                )
+                date_parts = re.search(r"Date=\"(.*)\"", header)
+                date_string_raw = date_parts.group(1) if date_parts else None
+                date_string = self._safe(
+                    lambda: parser.parse(date_string_raw, tzinfos={"GST": "UTC+4"})
+                )
+
+                report_date = epoch_time or date_string
+                if report_date:
+                    return {
+                        "collDate": report_date.astimezone(tz=timezone.utc)
+                        .isoformat(timespec="milliseconds")
+                        .replace("+00:00", "Z")
+                    }
+        return {}
+
     def create_manifest(self) -> dict[str, Any]:
         patient = self.__fetch_patient_by_kit_id()
 
@@ -167,6 +204,7 @@ class ManifestAssembler:
         return {
             "name": "Nebula",
             "indexedDate": datetime.now().strftime("%Y-%m-%d"),
+            **self.__extract_collection_date(),
             "reference": "GRCh38",
             "patientId": patient.get("id"),
             "mrn": self.__extract_elation_mrn(patient),

{phc-ingestion-0.10.12 → phc-ingestion-0.10.14}/ingestion/nebula/process.py

@@ -15,8 +15,9 @@ def process(vcf_file, out_path, file_name, source_file_id, ingestion_id, account
         )
 
         case_id = file_name
-        manifest_assembler = ManifestAssembler(ingestion_id, account_id, project_id, case_id)
-        manifest = manifest_assembler.create_manifest()
+        manifest = ManifestAssembler(
+            ingestion_id, account_id, project_id, case_id, vcf_file
+        ).create_manifest()
         base_vcf_file = os.path.basename(vcf_file)
         vcf_out = base_vcf_file.replace(".vcf", ".modified.vcf")
         vcf_final = base_vcf_file.replace(".vcf", ".modified.nrm.filtered.vcf")

{phc-ingestion-0.10.12 → phc-ingestion-0.10.14}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "phc-ingestion"
-version = "0.10.12"
+version = "0.10.14"
 description = "Functions for LifeOmic PHC genomic ingestions"
 authors = [
     { name = "LifeOmic Development", email = "development@lifeomic.com" },