phc-ingestion 0.9.2__py3-none-any.whl → 0.9.3__py3-none-any.whl

ingestion/nebula/__init__.py

@@ -0,0 +1 @@
+from ingestion.nebula.process import process

ingestion/nebula/process.py

@@ -0,0 +1,46 @@
+import os
+from pathlib import Path
+from ingestion.vcf_standardization.standardize import standardize_vcf
+from lifeomic_logging import scoped_logger
+
+
+def process(vcf_file: str, source_file_id: str, out_path: str, case_id: str) -> dict:
+    with scoped_logger(__name__) as log:
+        # TODO: do we need to take in + process the manifest file here?
+
+        # Process VCF
+        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")
+        if not vcf_final.endswith(".gz"):
+            vcf_final = vcf_final + ".gz"
+
+        # Assuming Nebula VCFs are germline
+        sample_name = f"germline_{case_id}"
+        vcf_line_count = standardize_vcf(
+            vcf_file, vcf_out, out_path, sample_name, log, compression=True
+        )
+
+        # Create a basic manifest for the Nebula VCF
+        manifest = {
+            "testType": "Nebula",
+            "sourceFileId": source_file_id,
+            "reference": "GRCh38",  # Assuming GRCh38, adjust as needed
+            "resources": [{"fileName": f".lifeomic/nebula/{case_id}/{base_vcf_file}"}],
+            "files": [
+                {
+                    "fileName": f".lifeomic/nebula/{case_id}/{vcf_final}",
+                    "sequenceType": "germline",
+                    "type": "shortVariant",
+                }
+            ],
+        }
+
+        case_metadata = {
+            "test_type": "Nebula",
+            "vcf_line_count": vcf_line_count,
+            "case_id": case_id,
+            "germline_genome_reference": manifest["reference"],
+        }
+
+        return case_metadata, manifest

{phc_ingestion-0.9.2.dist-info → phc_ingestion-0.9.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phc-ingestion
-Version: 0.9.2
+Version: 0.9.3
 Summary: Functions for LifeOmic PHC genomic ingestions
 License: MIT
 Author-email: LifeOmic Development <development@lifeomic.com>

{phc_ingestion-0.9.2.dist-info → phc_ingestion-0.9.3.dist-info}/RECORD

@@ -28,6 +28,8 @@ ingestion/foundation/util/vcf_etl.py,sha256=ZBrX1XGRz-ymLUEiVcjjqmPZPb-AfD9On8Uk
 ingestion/generic/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ingestion/generic/process.py,sha256=ZaVnZ_gx9faDUsuresI1A0oCegTa-dPQT7DBFMeZGyY,1777
 ingestion/generic/utils.py,sha256=1MEIru7uq38IjUdL8lcHqDH0oTki9uWrz1f2e-pmRoU,2814
+ingestion/nebula/__init__.py,sha256=VauK-rup_N8ZXVohx3HYqHX_PE_WoPyMUhdv2R7al4o,45
+ingestion/nebula/process.py,sha256=Ss2hReq9dAYFcBC4r89RYKcdE6cdUNXvgH2VMBijJPM,1687
 ingestion/nextgen/__init__.py,sha256=7LQ-h_Bvc5P1QcHMdzsqi1Qm4fTJn04-ozar2ty9wSc,59
 ingestion/nextgen/process.py,sha256=5Z0RfclwTAYZruGDiLPutjPCYFh1DJpoWY9dnttghT4,3993
 ingestion/nextgen/util/alteration_table.py,sha256=JTWBL1Fqj_pGsH5vwuVEnCUJle2wOBk6VYImHYCF9vg,6129
@@ -54,6 +56,6 @@ 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-0.9.2.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
-phc_ingestion-0.9.2.dist-info/METADATA,sha256=P8Wp02X90Sv3NntIt4K6h0hQNT0QrY1OpqDXaFTH-bg,572
-phc_ingestion-0.9.2.dist-info/RECORD,,
+phc_ingestion-0.9.3.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
+phc_ingestion-0.9.3.dist-info/METADATA,sha256=BeKjYNu_HC4PvcMS57E38mys7Q2gvpcrDTvaI0vvdPs,572
+phc_ingestion-0.9.3.dist-info/RECORD,,