phc-ingestion 0.8.35__py3-none-any.whl → 0.8.37__py3-none-any.whl

ingestion/nextgen/process.py

@@ -68,6 +68,7 @@ def process(
             bool(structural_path_name),
             translocations,
             hyperdiploidy_chromosomes,
+            log,
         )
         pre_filtered_somatic_vcf_path = pre_filter_somatic_vcf(
             vendor_files["somaticVcfFile"],

ingestion/nextgen/util/interpretation.py

@@ -18,11 +18,12 @@ def map_vendsig(vendsig: str) -> str:
     """
     Map vendor significance for short variants
     """
-    if vendsig in ["Pathogenic"]:
+    vendsig_lower = vendsig.lower()
+    if vendsig_lower in ["pathogenic"]:
         return "VENDSIG=Pathogenic"
-    elif vendsig in ["Likely Pathogenic", "LikelyPathogenic"]:
+    elif vendsig_lower in ["likely pathogenic", "likelypathogenic"]:
         return "VENDSIG=Likely pathogenic"
-    elif vendsig in ["VUS"]:
+    elif vendsig_lower in ["vus"]:
         return "VENDSIG=Uncertain significance"
     else:
         raise RuntimeError(f"Unable to map vendor significance: {vendsig}")

ingestion/nextgen/util/manifest_helpers.py

@@ -36,7 +36,53 @@ def parse_report_id(line: str) -> str:
     return parse_pattern(r"^.*Accession #: (.*?) .*$", line, "report ID")
 
 
-def parse_report_date(line: str) -> str:
+def parse_report_date_single_line(line: str) -> str:
     return parse_pattern(
         r"^.*Diagnostic Genomics Laboratory.*(\d{2}\/\d{2}\/\d{4}).*$", line, "report date"
     )
+
+
+def parse_report_date_multiline(patient_info_lines: list[str]) -> str:
+    in_range_trigger = False
+
+    for line in patient_info_lines:
+        if "Laboratory" in line:
+            in_range_trigger = True
+            continue
+        if in_range_trigger:
+            formatted_line = re.sub(r"<\/?T.\/?>", "", line).strip()
+            if not formatted_line:
+                continue
+            return parse_pattern(
+                r"^.*(\d{2}\/\d{2}\/\d{4}).*$", formatted_line, "report date from multiline"
+            )
+
+    raise ValueError("Could not parse report date from lines")
+
+
+def parse_report_date(patient_info_lines: list[str], log: Logger) -> str:
+    """
+    Typically, the report date is in a form like:
+    ```
+    Diagnostic Genomics Laboratory 01/01/2021
+    ```
+
+    However, sometimes the date is split across multiple lines, like:
+    ```
+    Diagnostic Genomics Laboratory
+    ...random empty lines or lines with only tags...
+    01/01/2021
+    ```
+    This function attempts to first parse the date from a single line, and if that fails,
+    it will attempt to parse it from multiple lines.
+    """
+    for line in patient_info_lines:
+        if "Laboratory" in line:
+            try:
+                report_date = parse_report_date_single_line(line)
+                return report_date
+            except ValueError:
+                log.warning("Could not parse report date from single line")
+                break
+
+    return parse_report_date_multiline(patient_info_lines)

ingestion/nextgen/util/process_manifest.py

@@ -17,7 +17,7 @@ def search_and_grab(array: list, search_item: str, grab_index: int):
     return array[array.index([i for i in array if re.search(search_item, i)][0]) + grab_index]
 
 
-def extract_xml_text(xml_in_file: str):
+def extract_xml_text(xml_in_file: str) -> list[str]:
     with open(xml_in_file, "r") as f:
         xml_lines = f.readlines()
 
@@ -35,7 +35,7 @@ def extract_xml_text(xml_in_file: str):
     return patient_info_lines
 
 
-def extract_interpretation_text(xml_in_file: str):
+def extract_interpretation_text(xml_in_file: str) -> list[str]:
     with open(xml_in_file, "r") as f:
         xml_lines = f.readlines()
 
@@ -111,7 +111,7 @@ def extract_patient_data(patient_info_lines: list[str]):
     return patient_data
 
 
-def extract_test_data(patient_info_lines: list, interpretation_lines: list):
+def extract_test_data(patient_info_lines: list[str], interpretation_lines: list[str], log: Logger):
     # Initialize manifest and hard-code some values
     manifest: dict[str, Any] = {}
     manifest["testType"] = "Plasma Cell Myeloma Panel"
@@ -128,12 +128,11 @@ def extract_test_data(patient_info_lines: list, interpretation_lines: list):
     manifest["medFacilID"] = ""
     manifest["medFacilName"] = "IU Health"
 
-    for line in patient_info_lines:
-        if "reportDate" not in manifest and "Laboratory" in line:
-            report_date = manifest_helpers.parse_report_date(line)
-            manifest["reportDate"] = transform_date(report_date)
-            manifest["indexedDate"] = manifest["reportDate"]
+    report_date = manifest_helpers.parse_report_date(patient_info_lines, log)
+    manifest["reportDate"] = transform_date(report_date)
+    manifest["indexedDate"] = manifest["reportDate"]
 
+    for line in patient_info_lines:
         if "collDate" not in manifest and "Collected" in line:
             collArray = line.split(" ")
             coll_date = search_and_grab(collArray, "Collected", 1)
@@ -178,10 +177,11 @@ def process_manifest(
     include_structural: bool,
     somatic_translocations: list[str],
     hyperdiploidy_chromosomes: list[str] | None,
+    log: Logger,
 ):
     test_text = extract_xml_text(xml_in_file)
     interpretation_text = extract_interpretation_text(xml_in_file)
-    manifest = extract_test_data(test_text, interpretation_text)
+    manifest = extract_test_data(test_text, interpretation_text, log)
     manifest.update(extract_patient_data(test_text))
 
     file_prefix = f".lifeomic/nextgen/{case_id}/{case_id}"

{phc_ingestion-0.8.35.dist-info → phc_ingestion-0.8.37.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phc-ingestion
-Version: 0.8.35
+Version: 0.8.37
 Summary: Functions for LifeOmic PHC genomic ingestions
 License: MIT
 Author-email: LifeOmic Development <development@lifeomic.com>

{phc_ingestion-0.8.35.dist-info → phc_ingestion-0.8.37.dist-info}/RECORD

@@ -29,14 +29,14 @@ ingestion/generic/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU
 ingestion/generic/process.py,sha256=WJHV_-SKhrDZ3JS3fm9DVMoW3Zs2t50GiraSV3vlLHE,1548
 ingestion/generic/utils.py,sha256=1MEIru7uq38IjUdL8lcHqDH0oTki9uWrz1f2e-pmRoU,2814
 ingestion/nextgen/__init__.py,sha256=7LQ-h_Bvc5P1QcHMdzsqi1Qm4fTJn04-ozar2ty9wSc,59
-ingestion/nextgen/process.py,sha256=kDCnU685v7aqJ3i4HpFdb7HqgHRSBKqtYPpuyN7qWmM,3976
+ingestion/nextgen/process.py,sha256=5Z0RfclwTAYZruGDiLPutjPCYFh1DJpoWY9dnttghT4,3993
 ingestion/nextgen/util/alteration_table.py,sha256=bvTrXEhIRye5BzSXjZEBy1AvXLZgG0rNkNt3e3rcvv0,6127
-ingestion/nextgen/util/interpretation.py,sha256=tQ3qzAUwBzwK1tQzH9ujZk_VCQ2wP8HzSZY0fImJ5Wo,818
-ingestion/nextgen/util/manifest_helpers.py,sha256=2xrpEtHbCb1Kea1wJeObkDfTiBklmffQt_o2hMgOSOE,1208
+ingestion/nextgen/util/interpretation.py,sha256=56wVk9j-w59gM-11iODXbKuUtZcEwe8zJQSXpjyCguw,872
+ingestion/nextgen/util/manifest_helpers.py,sha256=LH5em0xsu9Hrs175vfx6SX8W1Ww2FFRp2wIBSfIEMUM,2725
 ingestion/nextgen/util/nextgen_specific_genes.py,sha256=1jFcqvtYAlJ7eBwOBm1UC2TzAbjHjdlvPBUzxr1G8dY,1206
 ingestion/nextgen/util/pre_filter_somatic_vcf.py,sha256=mIaUihmGLbS38D4Gy_Qtf1lFAfW0A-LgAgQmsrEiI-M,3529
 ingestion/nextgen/util/process_cnv.py,sha256=MIirc8e0k6lsaTZkRM3U3L3IvbrcHmKQ4xlIu585514,2430
-ingestion/nextgen/util/process_manifest.py,sha256=EnV9I90vnanDvuoErbMfz6yAfjzM5LdhhUF4q5DJd8w,8428
+ingestion/nextgen/util/process_manifest.py,sha256=RsHzDGL1OBea2raoEHACo8owRodIFpX3xVE-aFOoyrg,8428
 ingestion/nextgen/util/process_structural.py,sha256=FKjkK7BkIlocnLs8rFCjrMC39FCQnD0nQCeWvi7cRoA,7539
 ingestion/nextgen/util/process_vcf.py,sha256=ZZURSMnZhHDpFahzijZ4MvCfSWTPdIktzmnCKVVUbGs,7768
 ingestion/nextgen/util/types.py,sha256=SSzt5gv-kss1PR45eQUelypWrGI-dAfQMO3GSD-T-Wg,22
@@ -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-0.8.35.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
-phc_ingestion-0.8.35.dist-info/METADATA,sha256=PctzE287ytgM7HHKcZO5VnMScgiX3bgpmgPNpF-ntWI,552
-phc_ingestion-0.8.35.dist-info/RECORD,,
+phc_ingestion-0.8.37.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
+phc_ingestion-0.8.37.dist-info/METADATA,sha256=06qKld0MIJchM-NfUQKojMeKDo1nnYHXPDpwLrbcJLk,552
+phc_ingestion-0.8.37.dist-info/RECORD,,