phc-ingestion 0.11.1__py3-none-any.whl → 1.0.1__py3-none-any.whl

ingestion/generic/utils.py

@@ -44,6 +44,7 @@ def check_manifest(manifest_file_path: str, case_id, log: Logger):
             Optional("plasmaTumorFraction"): str,
             Optional("cellPurity"): float,
             Optional("hrdStatus"): str,
+            Optional("sampleId"): str,
         }
     )
 

ingestion/vcf_standardization/util/read_write.py

@@ -4,6 +4,8 @@ import re
 import os
 from typing import Iterator, Optional
 
+import pysam
+
 
 def check_vcf(infile: str, log: Logger) -> None:
     log.info("Checking VCF file")
@@ -72,11 +74,35 @@ def write_vcf(
 ) -> int:
     log.info(f"Writing standardized VCF to {outfile}")
 
-    with gzip.open(outfile, "wt") if compression else open(outfile, "w") as w:
-        w.write("\n".join(headers) + "\n")
-        for variant in variants_gen:
-            line_count += 1
-            if variant:
-                w.write(variant + "\n")
+    if compression:
+        # Write uncompressed first, then compress with bgzip (indexable format)
+        # pysam.tabix_compress requires an uncompressed input file
+        # Create a temporary uncompressed file
+        if outfile.endswith(".gz"):
+            temp_uncompressed = outfile[:-3]  # Remove .gz extension
+        else:
+            temp_uncompressed = outfile + ".tmp"
+
+        with open(temp_uncompressed, "w") as w:
+            w.write("\n".join(headers) + "\n")
+            for variant in variants_gen:
+                line_count += 1
+                if variant:
+                    w.write(variant + "\n")
+
+        # Compress with bgzip (indexable format) using pysam
+        log.info(f"Compressing VCF with bgzip to {outfile}")
+        pysam.tabix_compress(temp_uncompressed, outfile, force=True)
+
+        # Clean up temporary uncompressed file
+        if os.path.exists(temp_uncompressed):
+            os.remove(temp_uncompressed)
+    else:
+        with open(outfile, "w") as w:
+            w.write("\n".join(headers) + "\n")
+            for variant in variants_gen:
+                line_count += 1
+                if variant:
+                    w.write(variant + "\n")
 
     return line_count

{phc_ingestion-0.11.1.dist-info → phc_ingestion-1.0.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phc-ingestion
-Version: 0.11.1
+Version: 1.0.1
 Summary: Functions for LifeOmic PHC genomic ingestions
 License: MIT
 Author-email: LifeOmic Development <development@lifeomic.com>
@@ -13,6 +13,7 @@ Requires-Dist: natsort==7.1.1
 Requires-Dist: numpy>=2.1.2
 Requires-Dist: packaging>=23.1
 Requires-Dist: pandas>=2.2.3
+Requires-Dist: pysam>=0.21.0
 Requires-Dist: ruamel.yaml==0.17.21
 Requires-Dist: schema>=0.7.5
 Requires-Dist: urllib3>=1.26.16

{phc_ingestion-0.11.1.dist-info → phc_ingestion-1.0.1.dist-info}/RECORD

@@ -25,7 +25,7 @@ ingestion/foundation/util/interpretation.py,sha256=LVVUmMyD6Un1rIKXqiyQDUC6oIJUd
 ingestion/foundation/util/vcf_etl.py,sha256=GXV5JXswwdyHEEdPsM3Qq8tDPFkvZajrZn5chWgF53k,2266
 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/generic/utils.py,sha256=YHzbmB0tP-aQjvY9vRSn8iOhailleYwvctDbB0IEKb8,2853
 ingestion/nebula/__init__.py,sha256=VauK-rup_N8ZXVohx3HYqHX_PE_WoPyMUhdv2R7al4o,45
 ingestion/nebula/constants.py,sha256=thKqSwemdaAwAmKvF4FEVI9l1Ph5ergsnMlx6nWte7E,357
 ingestion/nebula/manifest_assembler.py,sha256=kcRSy6pixHkuVEK9QSoM-i6ZdLWMSYXw39eKGHvam34,7995
@@ -59,7 +59,7 @@ ingestion/vcf_standardization/standardize.py,sha256=zYzZxncq8USA1bUs26L-ByLPTnUl
 ingestion/vcf_standardization/util/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 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-0.11.1.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
-phc_ingestion-0.11.1.dist-info/METADATA,sha256=ucEZCziomM9ANPMV8jqSBT5Huigd0UownHOnSKWqKdc,677
-phc_ingestion-0.11.1.dist-info/RECORD,,
+ingestion/vcf_standardization/util/read_write.py,sha256=xogLdqtm1xGzigY459LqP_1zM6c5X9AjAFGkfaDI-bg,3479
+phc_ingestion-1.0.1.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
+phc_ingestion-1.0.1.dist-info/METADATA,sha256=jvycAlRde4lUsP8O87F9a5bSF3fFINgtkMrr5FKsqWE,705
+phc_ingestion-1.0.1.dist-info/RECORD,,