filterCNVs: filterCNVs

In jpuntomarcos/CNVfilteR: Identifies false positives of CNV calling tools by using SNV calls

filterCNVs

Description

Identifies those copy number calls that can be filtered out

Usage

filterCNVs(
  cnvs.gr,
  vcfs,
  expected.ht.mean = 50,
  expected.dup.ht.mean1 = 33.3,
  expected.dup.ht.mean2 = 66.6,
  sigmoid.c1 = 2,
  sigmoid.c2.vector = c(28, 38.3, 44.7, 55.3, 61.3, 71.3),
  dup.threshold.score = 0.5,
  ht.deletions.threshold = 30,
  verbose = FALSE,
  margin.pct = 10
)

Arguments

cnvs.gr	GRanges containing CNVs to be filtered out. Use loadCNVcalls to load them.
vcfs	List of GRanges containing all variants (SNV/indel) obtaining with the loadVCFs function.
expected.ht.mean	Expected heterozygous SNV/indel allele frequency (defaults to 50)
expected.dup.ht.mean1	Expected heterozygous SNV/indel allele frequency when the variant IS NOT in the same allele than the CNV duplication call. (defaults to 33.3)
expected.dup.ht.mean2	Expected heterozygous SNV/indel allele frequency when the variant IS in the same allele than the CNV duplication call. (defaults to 66.6)
sigmoid.c1	Sigmoid c1 parameter. (defaults to 2)
sigmoid.c2.vector	Vector containing sigmoid c2 parameters for the six sigmoids functions. (defaults to c(28, 38.3, 44.7, 55.3, 61.3, 71.3))
dup.threshold.score	Limit value to decide if a CNV duplication can be filtered out or not. A CNV duplication can be filtered out if the total score computed from heterozygous variants in the CNV is equal or greater than dup.threshold.score. (defaults to 0.5)
ht.deletions.threshold	Minimum percentage of heterozygous variants falling in a CNV deletion to filter that CNV. (defaults to 30)
verbose	Whether to show information messages. (defaults to TRUE)
margin.pct	Variants in the CNV but close to the ends of the CNV will be ignored. margin.pct defines the percentage of CNV length, located at each CNV limit, where variants will be ignored. For example, for a CNV chr1:1000-2000 and a margin.pct value of 10, variants within chr1:1000-1100 and chr1:1900-2000 will be ignored.

Details

Checks all the variants (SNV and optionally INDELs) in each CNV present in cnvs.gr to decide whether a CNV can be filtered out or not. It returns an S3 object with 3 elments: cnvs, variantsForEachCNV and filterParameters. See return section for further details.

A CNV deletion can be filtered out if there is at least ht.deletions.threshold A CNV duplication can be filtered out if the score is >= dup.threshold.score after computing all heterozygous variants falling in the CNV.

If a CNV can be filtered out, then the value TRUE is set in the filter column of the cnvs element.

Value

A S3 object with 3 elements:

• cnvs: GRanges with the input CNVs and the meta-columns added during the call:

  • cnv.id: CNV id

  • filter: Set to TRUE if the CNV can be filtered out

  • n.total.variants: Number of variants in the CNV

  • n.hm.variants: Number of homozygous variants. They do not give any evidenced for confirming or discarding the CNV.

  • n.ht.discard.CNV: For a CNV duplication, number of heterozygous variants in that discard the CNV (those with a positive score)

  • n.ht.confirm.CNV: For a CNV duplication, number of heterozygous variants that confirm the CNV (those with a negative score)

  • ht.pct: Percentage of heterozygous variants for deletion CNVs

  • score: total score when computing all the variants scores

• variantsForEachCNV: named list where each name correspond to a CNV id and the value is a data.frame with all variants falling in that CNV

• filterParameters: input parameters used for filtering

Examples

# Load CNVs data
cnvs.file <- system.file("extdata", "DECoN.CNVcalls.csv", package = "CNVfilteR", mustWork = TRUE)
cnvs.gr <- loadCNVcalls(cnvs.file = cnvs.file, chr.column = "Chromosome", start.column = "Start", end.column = "End", cnv.column = "CNV.type", sample.column = "Sample")

# Load VCFs data
vcf.files <- c(system.file("extdata", "variants.sample1.vcf.gz", package = "CNVfilteR", mustWork = TRUE),
               system.file("extdata", "variants.sample2.vcf.gz", package = "CNVfilteR", mustWork = TRUE))
vcfs <- loadVCFs(vcf.files, cnvs.gr = cnvs.gr)

# Filter CNVs
results <- filterCNVs(cnvs.gr, vcfs)

# Check CNVs that can be filtered out
as.data.frame(results$cnvs[results$cnvs$filter == TRUE])

jpuntomarcos/CNVfilteR documentation built on Oct. 6, 2023, 12:37 a.m.