AllelicSeries (version 0.1.1.5)

COAST: COding-variant Allelic Series Test

Description

Main allelic series test. Performs both Burden and SKAT type tests, then combines the results to calculate an omnibus p-value.

Usage

COAST(
  anno,
  geno,
  pheno,
  add_intercept = TRUE,
  apply_int = TRUE,
  covar = NULL,
  include_orig_skato_all = FALSE,
  include_orig_skato_ptv = FALSE,
  is_pheno_binary = FALSE,
  min_mac = 0,
  ptv_anno = 3,
  pval_weights = NULL,
  return_omni_only = FALSE,
  score_test = FALSE,
  weights = c(1, 2, 3)
)

Value

An object of class COAST with slots for effect sizes, variant counts, and p-values.

Arguments

anno

(snps x 1) annotation vector with integer values in 1 through the number of annotation categories L.

geno

(n x snps) genotype matrix.

pheno

(n x 1) phenotype vector.

add_intercept

Add an intercept if not present in the supplied covariate matrix covar? Default: TRUE.

apply_int

Apply rank-based inverse normal transform to the phenotype? Default: TRUE. Ignored if phenotype is binary.

covar

(n x p) covariate matrix. Defaults to an (n x 1) intercept.

include_orig_skato_all

Include the original version of SKAT-O applied to all variants in the omnibus test? Default: FALSE.

include_orig_skato_ptv

Include the original version of SKAT-O applied to PTV variants only in the omnibus test? Default: FALSE.

is_pheno_binary

Is the phenotype binary? Default: FALSE.

min_mac

Minimum minor allele count for inclusion. Default: 0.

ptv_anno

Annotation of the PTV category, only required if include_orig_skato_ptv is set to TRUE.

pval_weights

Optional vector of relative weights for combining the component tests to perform the omnibus test. By default, 50% of weight is given to the 6 burden tests, and 50% to the 1 SKAT test. If specified, the weight vector should have length 7, and the length should be increased if either include_orig_skato_all or include_orig_skato_ptv is active.

return_omni_only

Return only the omnibus p-value? Default: FALSE.

score_test

Use a score test for burden analysis? If FALSE, uses a Wald test.

weights

(L x 1) vector of annotation category weights. Note that the number of annotation categories L is inferred from the length of weights.

Examples

# Generate data.
data <- DGP(n = 1e3, snps = 1e2)

# Run the COding-variant Allelic Series Test.
results <- COAST(
  anno = data$anno,
  geno = data$geno,
  pheno = data$pheno,
  covar = data$covar
)
show(results)