read_vcf_cpp: Read VCF using CPP reader
In jmonlong/sveval: SV evaluation

read_vcf_cpp

R Documentation

Read VCF using CPP reader

Description

For each VCF record the information in the INFO field is used in priority. If missing, information is guessed from the REF/ALT sequences. If multiple alleles are defined in ALT, they are split and the allele count extracted from the GT field.

Usage

read_vcf_cpp(
  filename,
  use_gz,
  sample_name = "",
  min_sv_size = 10L,
  shorten_ref = TRUE,
  shorten_alt = TRUE,
  gq_field = "GQ",
  check_inv = FALSE,
  keep_nocalls = FALSE,
  other_fields = as.character(c())
)

Arguments

`filename`	the path to the VCF file (unzipped or gzipped).
`use_gz`	is the VCF file gzipped?
`sample_name`	which sample to process. If not found, uses first sample in VCF file. If "*", force no sample selection
`min_sv_size`	minimum variant size to keep in bp. Variants shorter than this will be skipped. Default is 10.
`shorten_ref`	should the REF sequence be shortened to the first 10 bp. Default is TRUE
`shorten_alt`	should the ALT sequence be shortened to the first 10 bp. Default is TRUE
`gq_field`	which field from FORMAT should be used as genotype quality. Default is "GQ". If not found, QUAL will be used
`check_inv`	guess if a variant is an inversion by aligning REF with the reverse complement of ALT. If >80% similar (and REF and ALT>10bp), variant is classified as INV.
`keep_nocalls`	should we keep variants/alleles with missing genotypes (e.g. "./."). Default is FALSE
`other_fields`	name of another field from INFO to extract.

Details

Alleles are split and, for each, column 'ac' reports the allele count. Notable cases incude 'ac=-1' for no/missing calls (e.g. './.'), and 'ac=0' on the first allele to report hom ref, variants. These cases are often filtered later with 'ac>0' to keep only non-ref calls. If the VCF contains no samples or if no sample selection if forced (sample_name='*'), 'ac' will contain '-1' for all variants in the VCF.