Fill empty ANN values with NA to ensure a correct extraction

Author: ahmig

workflow/rules/vaf.smk

@@ -160,7 +160,7 @@ rule extract_vcf_fields:
     log:
         LOGDIR / "tsv_to_vcf" / "{sample}.log.txt"
     shell:
-        'SnpSift extractFields -s {params.sep:q} {input.vcf:q} {params.extract_columns} >{output.tsv:q} 2>{log:q}'
+        "SnpSift extractFields -e 'NA' -s {params.sep:q} {input.vcf:q} {params.extract_columns} >{output.tsv:q} 2>{log:q}"
 
 
 rule format_vcf_fields_longer:

workflow/scripts/format_vcf_fields_longer.R

@@ -19,9 +19,7 @@ filter.exclude <- lapply(snakemake@params$filter_exclude, empty.to.na)
 # Process input table
 read_tsv(snakemake@input$tsv) %>%
     # Separate <sep>-delimited "...[*]..." columns (e.g. ANN[*].EFFECT)
-    separate_longer_delim(contains("[*]"), delim = snakemake@params$sep) %>%
-    # Replace empty ("") fields with NA
-    mutate(across(contains("[*]"), ~ na_if(., ""))) %>%
+    separate_rows(contains("[*]"), sep = snakemake@params$sep, convert = TRUE) %>%
     # Rename "...[*]..." columns using the provided lookup via Snakemake config
     rename(all_of(unlist(snakemake@params$colnames_mapping))) %>%
     # Apply dynamic filters from the Snakemake config: