phc-ingestion 0.8.25__py3-none-any.whl → 0.8.26__py3-none-any.whl

ingestion/nextgen/util/pre_filter_somatic_vcf.py

@@ -9,6 +9,17 @@ def build_variant_key_from_vcf_line(line: str) -> str:
     return f"{chrom}:{pos}:{ref}:{alt}"
 
 
+def extract_filter_from_vcf_line(line: str) -> str:
+    split_line = line.strip().split("\t")
+    return split_line[6]
+
+
+def replace_filter_in_vcf_line(line: str, new_filter: str) -> str:
+    split_line = line.strip().split("\t")
+    split_line[6] = new_filter
+    return "\t".join(split_line) + "\n"
+
+
 def pre_filter_somatic_vcf(
     somatic_vcf_file: str,
     somatic_vcf_snv_file: str,
@@ -19,22 +30,25 @@ def pre_filter_somatic_vcf(
     """
     Removes all variants from the `somatic_vcf_file` that are not
     also in the `somatic_vcf_snv_file` or `somatic_vcf_indel_file`.
+
+    Also updates the FILTER field in the `somatic_vcf_file` to match
+    the FILTER field of the corresponding variant in the
+    `somatic_vcf_snv_file` or `somatic_vcf_indel_file`.
     """
     log.info("Pre-filtering somatic VCF file")
 
-    valid_variant_keys = set()
-    with open_maybe_gzipped(somatic_vcf_snv_file, "rt") as f:
-        for line in f:
-            if line.startswith("#"):
-                continue
-            valid_variant_keys.add(build_variant_key_from_vcf_line(line))
-    with open_maybe_gzipped(somatic_vcf_indel_file, "rt") as f:
-        for line in f:
-            if line.startswith("#"):
-                continue
-            valid_variant_keys.add(build_variant_key_from_vcf_line(line))
+    valid_variants_with_filters: dict[str, str] = {}
+
+    for file in [somatic_vcf_snv_file, somatic_vcf_indel_file]:
+        with open_maybe_gzipped(file, "rt") as f:
+            for line in f:
+                if line.startswith("#"):
+                    continue
+                valid_variants_with_filters[build_variant_key_from_vcf_line(line)] = (
+                    extract_filter_from_vcf_line(line)
+                )
 
-    log.info(f"Found {len(valid_variant_keys)} valid variants")
+    log.info(f"Found {len(valid_variants_with_filters)} valid variants")
 
     output_vcf_path = f"{working_dir}/filtered_somatic.vcf.gz"
     with (
@@ -45,8 +59,9 @@ def pre_filter_somatic_vcf(
             if line.startswith("#"):
                 w.write(line)
             else:
-                if build_variant_key_from_vcf_line(line) in valid_variant_keys:
-                    w.write(line)
+                key = build_variant_key_from_vcf_line(line)
+                if key in valid_variants_with_filters:
+                    w.write(replace_filter_in_vcf_line(line, valid_variants_with_filters[key]))
 
     log.info(f"Successfully pre-filtered somatic VCF file to {output_vcf_path}")
     return output_vcf_path

{phc_ingestion-0.8.25.dist-info → phc_ingestion-0.8.26.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phc-ingestion
-Version: 0.8.25
+Version: 0.8.26
 Summary: Functions for LifeOmic PHC genomic ingestions
 License: MIT
 Author-email: LifeOmic Development <development@lifeomic.com>

{phc_ingestion-0.8.25.dist-info → phc_ingestion-0.8.26.dist-info}/RECORD

@@ -33,7 +33,7 @@ ingestion/nextgen/process.py,sha256=GKiQ2dCxrR7tBD8TSP6Wk-TyoX3xBGip5zfpeT2buiQ,
 ingestion/nextgen/util/alteration_table.py,sha256=KwpJCQv_rVsL30jkzgZn0bKdd205fjVodYBNTcK3D1s,4220
 ingestion/nextgen/util/interpretation.py,sha256=ozuzb0vozff34zfP6AdOiUmI8Q77hI02jve_nCPZHfE,297
 ingestion/nextgen/util/manifest_helpers.py,sha256=PpSay-pe62jk735nom1tVD9nDE8-CxmzzCrgpBhgtjY,1571
-ingestion/nextgen/util/pre_filter_somatic_vcf.py,sha256=beAtOpznVtUOWKCUNNv1UukNBSa0LKmhXsJ_3K5xk2E,1759
+ingestion/nextgen/util/pre_filter_somatic_vcf.py,sha256=K_gH4EnUXrKB22u_f8FqQVGrOS5LxXNsNO3VBn381eY,2301
 ingestion/nextgen/util/process_cnv.py,sha256=m-AhsXFlYw4LTzgJJaj5vXYbK5n3H7cImzBxD2To6M0,2598
 ingestion/nextgen/util/process_manifest.py,sha256=TAbCHwN_2m08jySn1J4qEd9Nrxjw9CrsspwFWack0V4,8448
 ingestion/nextgen/util/process_structural.py,sha256=BXhwbRtFLTZsDi4ioSna1qmMtxVcsB0R-xNIvymm5Vw,5947
@@ -53,6 +53,6 @@ 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=IQotJ27To1MoQcRstc5AbHZtUuJz5cqkkZiHsDNaBvI,2471
-phc_ingestion-0.8.25.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
-phc_ingestion-0.8.25.dist-info/METADATA,sha256=AU2E4U0eLh5aXCGiCmVsVloyruqmkF_oNdkv7rmAfBQ,552
-phc_ingestion-0.8.25.dist-info/RECORD,,
+phc_ingestion-0.8.26.dist-info/WHEEL,sha256=B19PGBCYhWaz2p_UjAoRVh767nYQfk14Sn4TpIZ-nfU,87
+phc_ingestion-0.8.26.dist-info/METADATA,sha256=e1BZcEIqQwMUSnYlBbGz4BLsHPYVZg8UOBOxbtxiZ-A,552
+phc_ingestion-0.8.26.dist-info/RECORD,,