phc-ingestion 1.0.1__py3-none-any.whl → 1.0.2__py3-none-any.whl

ingestion/vcf_standardization/util/read_write.py

@@ -4,8 +4,6 @@ import re
 import os
 from typing import Iterator, Optional
 
-import pysam
-
 
 def check_vcf(infile: str, log: Logger) -> None:
     log.info("Checking VCF file")
@@ -74,35 +72,11 @@ def write_vcf(
 ) -> int:
     log.info(f"Writing standardized VCF to {outfile}")
 
-    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")
+    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")
 
     return line_count

{phc_ingestion-1.0.1.dist-info → phc_ingestion-1.0.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phc-ingestion
-Version: 1.0.1
+Version: 1.0.2
 Summary: Functions for LifeOmic PHC genomic ingestions
 License: MIT
 Author-email: LifeOmic Development <development@lifeomic.com>
@@ -13,7 +13,6 @@ 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-1.0.1.dist-info → phc_ingestion-1.0.2.dist-info}/RECORD

@@ -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=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,,
+ingestion/vcf_standardization/util/read_write.py,sha256=x3Pf6Dq8tmolblbCS5CrNmrcHS3FGfqBSFpFgvFGC4g,2526
+phc_ingestion-1.0.2.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
+phc_ingestion-1.0.2.dist-info/METADATA,sha256=PrNukFYVStjbSRrBD7OjQTv0ZVOIZMwbQ2LDgWP6lNM,676
+phc_ingestion-1.0.2.dist-info/RECORD,,