MultiSKAT_NULL: Construct and get parameters and residuals from the null...

In diptavo/MultiSKAT: Sequence Kernel Association Tests with Multiple phenotypes

Description

Compute the null model parameters and residuals for MultiSKAT. The values and residuals from here are used to compute the MultiSKAT test.

Usage

MultiSKAT_NULL(y.mat, X,is.fast=TRUE,method = "nlminb")

Arguments

y.mat	A numeric matrix of phenotype values with rows as individuals and columns as phenotypes.
X	A numeric matrix of covariate values with rows as individuals and columns as covariates. The first column should be the intercept (a vector of 1's).
is.fast	A binary variable indicating whether a fast approximation should be used or not (default = TRUE).
method	Indicates the optimizer to use for the generalized least squares procedure (default = "nlminb". Other option is "optim"). Works only with is.fast = FALSE

Details

For most of the scenarios with high sample size (more than 2000), is.fast= TRUE should be used. For lower sample size, is.fast=FALSE should be set. There might be cases where the default optimizer "nlminb" fails to converge due to numerical issue. Users can switch to another optimzer "optim" in that case by setting method="optim".

Value

This function returns an null object of the class SKAT_Multiphen_Null_PhenCorr. This object contains details, values and residuals for computing MultiSKAT test statistic and the null distribution.

Author(s)

Diptavo Dutta

Examples

data(MultiSKAT.example.data)
attach(MultiSKAT.example.data)

obj.null <- MultiSKAT_NULL(Phenotypes,Cov)

diptavo/MultiSKAT documentation built on May 22, 2019, 1:36 p.m.