CKAT {CKAT}

Composite kernel association test for SNP-set analysis in pharmacogenetics (PGx) studies.

Description

Composite kernel association test for SNP-set analysis in pharmacogenetics (PGx) studies.

Usage

CKAT(G, Tr, X, y, trait = "continuous", ker = "linear", grids = c(0,
  0.5, 1), n_a = 1000, method = "liu", subdiv = 10^6)

Arguments

G

- genotype matrix.

Tr

- treatment vector, 0 indicates placebo, 1 indicates treatment.

X

- non-genetic covariates data matrix.

y

- response vector. Currently continuous and binary responses are supported. Survival response will be added soon.

trait

- response indicator. trait = "continuous" or "binary".

ker

- kernel. ker = "linear", "IBS", "Inter" (interaction kernel) and "RBF" (radial basis function kernel).

grids

- grids of the candidate weights.

n_a

- the number of intervals for manual integration (when integrate function fails). Default n_a = 1000.

method

- method for getting density of A (see details in the reference). Default method is Liu's method.

subdiv

- parameter of Davies' method. Default value is 1E6.

Value

pvals - p-values of each individual association test.

finalp - final p-value of the CKAT test.

Examples

nsamples = 500; nsnps = 10
X = rnorm(nsamples,0,1)
Tr = sample(0:1,nsamples,replace=TRUE)
G = matrix(rbinom(nsamples*nsnps, 1, 0.05), nrow = nsamples, ncol = nsnps)
GxT = G*Tr
Y0 = 0.5*X + Tr + rnorm(nsamples)
CKAT(G, Tr, X, Y0, grids=c(0,0.5,1))