phc-ingestion 0.8.42__py3-none-any.whl → 0.9.0__py3-none-any.whl

ingestion/caris/util/metadata.py

@@ -94,9 +94,13 @@ def extract_metadata(data, prefix, files, source_file_id, log: Logger) -> dict:
     # Get date of collected and received for the specimen
     metadata["receivedDate"] = specimen_details["receivedDate"]
     metadata["collDate"] = specimen_details["collDate"]
+    if specimen_details["specimenId"]:
+        metadata["specimenIds"] = [specimen_details["specimenId"]]
+    if specimen_details["specimenTypes"]:
+        metadata["specimenTypes"] = specimen_details["specimenTypes"]
     metadata["reportDate"] = get_report_date(test_details, log)
 
-    #  Get the date without the time
+    # Get the date without the time
     metadata["indexedDate"] = metadata["reportDate"]
 
     patient = data["patientInformation"]

ingestion/caris/util/specimen_details.py

@@ -1,5 +1,5 @@
 from logging import Logger
-from typing import TypedDict, cast
+from typing import TypedDict, cast, Optional
 
 
 class SpecimenDetails(TypedDict, total=False):
@@ -8,37 +8,55 @@ class SpecimenDetails(TypedDict, total=False):
     specimenReceivedDate: str
     specimenCollectionDate: str
     specimenSite: str
+    specimenType: str
+    specimenID: str
 
 
 class ParsedSpecimenDetails(TypedDict):
     bodySite: str
     receivedDate: str
     collDate: str
+    specimenId: Optional[str]
+    specimenTypes: Optional[list[str]]
 
 
-def parse_specimen_details(specimen_details: SpecimenDetails) -> ParsedSpecimenDetails:
+def parse_specimen_details(specimen_details: list[SpecimenDetails]) -> ParsedSpecimenDetails:
+    specimen_types: list[str] = []
+    for specimen in specimen_details:
+        specimen_type = specimen.get("specimenType")
+        if specimen_type and specimen_type not in specimen_types:
+            specimen_types.append(specimen_type)
+
     return {
-        "bodySite": specimen_details.get("specimenSite", ""),
-        "receivedDate": specimen_details.get("specimenReceivedDate", ""),
-        "collDate": specimen_details.get("specimenCollectionDate", ""),
+        "bodySite": specimen_details[0].get("specimenSite", ""),
+        "receivedDate": specimen_details[0].get("specimenReceivedDate", ""),
+        "collDate": specimen_details[0].get("specimenCollectionDate", ""),
+        "specimenId": specimen_details[0].get("specimenID"),
+        "specimenTypes": specimen_types if specimen_types else None,
     }
 
 
-def ensure_single_specimen_details(
+def standardize_specimen_details(
     specimen_details: SpecimenDetails | list[SpecimenDetails],
     log: Logger,
-) -> SpecimenDetails:
+) -> list[SpecimenDetails]:
+    """
+    Specimen details can be a single dictionary or a list of dictionaries
+    This function standardizes the input to always be a list of dictionaries
+
+    If it is a list, we expect all the specimens to have the same site and warn otherwise.
+    """
     if isinstance(specimen_details, dict):
-        return specimen_details
+        return [specimen_details]
 
     # Sometimes, we have multiple specimen details
     # In this case, we expect them to all be the same and warn otherwise
-    sites = {specimen["specimenSite"] for specimen in specimen_details}
+    sites = {specimen.get("specimenSite") for specimen in specimen_details}
 
     if len(sites) > 1:
-        log.warn(f"Multiple specimen sites found")
+        log.warning("Multiple specimen sites found")
 
-    return specimen_details[0]
+    return specimen_details
 
 
 def extract_and_parse_specimen_details(data: dict, log: Logger) -> ParsedSpecimenDetails:
@@ -62,6 +80,6 @@ def extract_and_parse_specimen_details(data: dict, log: Logger) -> ParsedSpecime
     if not specimen_details:
         raise ValueError("No specimen details found in data")
 
-    specimen_details = ensure_single_specimen_details(specimen_details, log)
+    specimen_details = standardize_specimen_details(specimen_details, log)
 
     return parse_specimen_details(specimen_details)

ingestion/foundation/util/ga4gh.py

@@ -190,6 +190,19 @@ def get_test_yml(
     reportDate = str(get_date(signatures.get("Signature").get("ServerTime")))[0:10]
     indexedDate = reportDate
 
+    sample_id = (sample.get("SampleId") or "").strip()
+    block_id = (sample.get("BlockId") or "").strip()
+    if block_id == "N/A":
+        block_id = ""
+
+    specimen_ids = []
+    if sample_id:
+        specimen_ids.append(sample_id)
+    if block_id:
+        specimen_ids.append(block_id)
+
+    specimenType = sample.get("SpecFormat", "").strip()
+
     yaml_file = {
         "name": "Foundation Medicine",
         "reference": "GRCh37",
@@ -216,6 +229,11 @@ def get_test_yml(
         "files": [],
     }
 
+    if specimen_ids:
+        yaml_file["specimenIds"] = specimen_ids
+    if specimenType:
+        yaml_file["specimenTypes"] = [specimenType]
+
     if write_to_manifest["cnv"]:
         yaml_file["files"].append(
             {

{phc_ingestion-0.8.42.dist-info → phc_ingestion-0.9.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phc-ingestion
-Version: 0.8.42
+Version: 0.9.0
 Summary: Functions for LifeOmic PHC genomic ingestions
 License: MIT
 Author-email: LifeOmic Development <development@lifeomic.com>

{phc_ingestion-0.8.42.dist-info → phc_ingestion-0.9.0.dist-info}/RECORD

@@ -9,8 +9,8 @@ ingestion/caris/util/hla.py,sha256=X_t6ngBRvmdG3m4I2_KnPFeWn3BaH-3IWHtOvDbS32A,7
 ingestion/caris/util/ihc.py,sha256=vegxudxHj7tLihrXGbEx_ptwkSsu3YCCB1nZVwoiYXg,12312
 ingestion/caris/util/interpretation.py,sha256=CghNurqeVA5VTBBorU8-ZTN-PVNPnR8wrmTwKCH3568,555
 ingestion/caris/util/json.py,sha256=HBU3Tf-XSi9fGHANYUtD8maXNYqmmnpncGh0KCDaPEU,5018
-ingestion/caris/util/metadata.py,sha256=C50e5a6zqYeUG_RcZvFvN-UEXWNJb0q03dMOGWkDgO0,10070
-ingestion/caris/util/specimen_details.py,sha256=R3uKHlLR056XcQbUPI6IO2dLr-z5Z5AJi866DJ379Qw,2105
+ingestion/caris/util/metadata.py,sha256=a6NToMtGtIRrlMd3CQwq4IRjGGmIiBA9JFwsATjNEoQ,10287
+ingestion/caris/util/specimen_details.py,sha256=5Tew0uhSa_SGBFEWjHp8wOKmQdeAYsY3-yTB-dEbtE4,2879
 ingestion/caris/util/structural.py,sha256=EUcMIea_WnafoVmFLIyEqlJ_HtYIj_g6qkekXa7QNQs,4628
 ingestion/caris/util/tar.py,sha256=BGR_2vBbxyMgF-GzJ3SrihsPdOzII4SFVz9tvKV5vo0,482
 ingestion/caris/util/tests.py,sha256=mcG3A8TW81_sn2Bfoa-Gd6Q1sR3_R4FX2BNskD4DkJk,372
@@ -22,7 +22,7 @@ ingestion/foundation/process.py,sha256=T8YTvXRiThqE1LTERhrzvvD69mP4qJ7soJ1ZIbu8Y
 ingestion/foundation/util/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ingestion/foundation/util/cnv.py,sha256=YSKCaOBhjZDNXth_GxC-50crDURpTNCMefoHo0uO7lQ,4215
 ingestion/foundation/util/fnv.py,sha256=-VstGsBKXM0duC-IpwUkektoTZ9yQUR0IQcDb1HibY0,5937
-ingestion/foundation/util/ga4gh.py,sha256=dn7uoQdnKHfIVCSHVEyPpR0jc1uRp1YB4P_9tLMOxMs,11134
+ingestion/foundation/util/ga4gh.py,sha256=R9qfFTjucOXntlPYq37g8-NYygk8HuCxz1vih_2iD24,11626
 ingestion/foundation/util/interpretation.py,sha256=LVVUmMyD6Un1rIKXqiyQDUC6oIJUd8cU3I9YHD5fsXg,405
 ingestion/foundation/util/vcf_etl.py,sha256=ZBrX1XGRz-ymLUEiVcjjqmPZPb-AfD9On8UkZJDa1Dk,2133
 ingestion/generic/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -54,6 +54,6 @@ ingestion/vcf_standardization/util/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQe
 ingestion/vcf_standardization/util/af_helpers.py,sha256=dpTzoeIQVeBRt0ETF3a9rp5ojZqznHg4x_hCZ8OPcOg,1061
 ingestion/vcf_standardization/util/dp_helpers.py,sha256=Nq8oLOLObu4_pv16qwwgpALRlUoJVCULrd9cFOD-eoI,823
 ingestion/vcf_standardization/util/read_write.py,sha256=x3Pf6Dq8tmolblbCS5CrNmrcHS3FGfqBSFpFgvFGC4g,2526
-phc_ingestion-0.8.42.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
-phc_ingestion-0.8.42.dist-info/METADATA,sha256=LB_E096gDykQT0yFc3_AP28VNff1Zi91Ltj971FZ8c0,573
-phc_ingestion-0.8.42.dist-info/RECORD,,
+phc_ingestion-0.9.0.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
+phc_ingestion-0.9.0.dist-info/METADATA,sha256=4sKbowdUIImMSp2a_3tm8WPyS7fq8hp-_96HRIZi_H0,572
+phc_ingestion-0.9.0.dist-info/RECORD,,