fix: treat ref and allele columns as non-VCF compliant

RIVM-bioinformatics · Jan 26, 2024 · 0e6a956 · 0e6a956
1 parent cd94fed
commit 0e6a956
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 9 deletions.
diff --git a/config/presets.yaml b/config/presets.yaml
@@ -2,7 +2,7 @@ mycobacterium_tuberculosis:
   single_copy_bed: files/mtb/coll_positions.bed
   reference_genbank: /mnt/db/juno/variant-typing/mtb/NC_000962.3.gbk
   resistance_variants_csv: /mnt/db/juno/variant-typing/mtb/reslist.csv
-  resistance_variants_columns: gene,drug,RIVM,TP,FP,confidence,variant_common_name
+  resistance_variants_columns: ref_nt,alt_nt,gene,drug,RIVM,TP,FP,confidence,variant_common_name
   resistance_variants_ab_column: drug
 other:
   single_copy_bed: null

diff --git a/workflow/rules/mtb_prepare_files.smk b/workflow/rules/mtb_prepare_files.smk
@@ -185,8 +185,6 @@ rule prepare_ab_table:
         ),
     params:
         POS="genomepos",
-        REF="ref",
-        ALT="allele",
         metadata=lambda wildcards: SAMPLES[wildcards.sample][
             "resistance_variants_columns"
         ],
@@ -199,10 +197,9 @@ rule prepare_ab_table:
 python workflow/scripts/convert_ab_table.py \
 --force-chrom NC_000962.3 \
 --POS {params.POS} \
---REF {params.REF} \
---ALT {params.ALT} \
 --other {params.metadata:q} \
-{input} {output.uncompressed} 2>&1>{log}
+{input} \
+{output.uncompressed} 2>&1>{log}
         """
 
 

diff --git a/workflow/scripts/convert_ab_table.py b/workflow/scripts/convert_ab_table.py
@@ -14,8 +14,8 @@
     type=str,
 )
 parser.add_argument("--POS", help="Column to parse genome positions", type=str)
-parser.add_argument("--REF", help="Column to parse reference allele", type=str)
-parser.add_argument("--ALT", help="Column to parse alternative allele", type=str)
+# parser.add_argument("--REF", help="Column to parse reference allele", type=str)
+# parser.add_argument("--ALT", help="Column to parse alternative allele", type=str)
 parser.add_argument(
     "--other",
     help="""
@@ -37,7 +37,8 @@
 
 metadata_cols = args.other.split(",")
 
-ordered_columns = [args.POS] + [args.REF] + [args.ALT] + metadata_cols
+# ordered_columns = [args.POS] + [args.REF] + [args.ALT] + metadata_cols
+ordered_columns = [args.POS] + metadata_cols
 
 for col in ordered_columns:
     assert col in df_in.columns, f"Column {col} cannot be found in {args.input}"