R: Filter VCF files

filterVcf {VariantAnnotation}

R Documentation

Filter VCF files

Description

Filter Variant Call Format (VCF) files from one file to another

Usage

## S4 method for signature 'character'
filterVcf(file, genome, destination, ..., verbose = TRUE,
    index = FALSE, prefilters = FilterRules(), filters = FilterRules(),
    param = ScanVcfParam())

## S4 method for signature 'TabixFile'
filterVcf(file, genome, destination, ..., verbose = TRUE,
    index = FALSE, prefilters = FilterRules(), filters = FilterRules(),
    param = ScanVcfParam())

Arguments

`file`	A `character(1)` file path or `TabixFile` specifying the VCF file to be filtered.
`genome`	A `character(1)` identifier
`destination`	A `character(1)` path to the location where the filtered VCF file will be written.
`...`	Additional arguments, possibly used by future methods.
`verbose`	A `logical(1)` indicating whether progress messages should be printed.
`index`	A `logical(1)` indicating whether the filtered file should be compressed and indexed (using `bgzip` and `indexTabix`).
`prefilters`	A `FilterRules` instance contains rules for filtering un-parsed lines of the VCF file.
`filters`	A `FilterRules` instance contains rules for filtering fully parsed VCF objects.
`param`	A `ScanVcfParam` instance restricting input of particular `info` or `geno` fields, or genomic locations. Applicable when applying a `filter` only. Prefiltering involves a grep of unparsed lines in the file; indexing is not used.

Details

This function transfers content of one VCF file to another, removing records that fail to satisfy prefilters and filters. Filtering is done in a memory efficient manner, iterating over the input VCF file in chunks of default size 100,000 (when invoked with character(1) for file) or as specified by the yieldSize argument of TabixFile (when invoked with TabixFile).

There are up to two passes. In the first pass, unparsed lines are passed to prefilters for filtering, e.g., searching for a fixed character string. In the second pass lines successfully passing prefilters are parsed into VCF instances and made available for further filtering. One or both of prefilter and filter can be present.

Filtering works by removing the rows (variants) that do not meet a criteria. Because this is a row-based approach and samples are column-based most genotype filters are only meaningful for single-sample files. If a single samples fails the criteria the entire row (all samples) are removed. The case where genotype filtering is effective for multiple samples is when the criteria is applied across samples and not to the individual (e.g., keep rows where all samples have DP > 10).

Value

The destination file path as a character(1).

Author(s)

Martin Morgan mailto:mtmorgan@fhcrc.org and Paul Shannon mailto:pshannon@fhcrc.org.

Examples

fl <- system.file(package="VariantAnnotation", "extdata",
                  "chr22.vcf.gz")
destination <- tempfile()
pre <- FilterRules(list(isLowCoverageExomeSnp = function(x) {
    grepl("LOWCOV,EXOME", x, fixed=TRUE)
}))
filt <- FilterRules(list(isSNP = function(x) info(x)$VT == "SNP"))
filtered <- filterVcf(fl, "hg19", destination, prefilters=pre, filters=filt)
vcf <- readVcf(filtered, "hg19")

[Package VariantAnnotation version 1.14.6 Index]