SKAT.example: Data example from SKAT package

In bigQF: Quadratic Forms in Large Matrices

Description

These data (probably synthetic) come from the SKAT package. The data set is too small for the leading-eigenvalue approximation to really make sense, but it provides some numerical comparison. The SKAT Q statistic should match exactly, the p-values should be fairly close.

Usage

data("SKAT.example")

Format

SKAT.example contains the following objects:

Z

a numeric genotype matrix of 2000 individuals and 67 SNPs. Each row represents a different individual, and each column represents a different SNP marker.

X

a numeric matrix of 2 covariates.

y.c

a numeric vector of continuous phenotypes.

y.b

a numeric vector of binary phenotypes.

Source

https://www.hsph.harvard.edu/skat/

Examples

data(SKAT.example)

skat1mf <- SKAT.matrixfree(SKAT.example$Z)
Q<-skat1mf$Q(SKAT.example$y.c)
all.equal(as.numeric(Q), 234803.786)
## correct value is 0.01874576
pQF(Q, skat1mf, neig=4, convolution.method="integration")

skat2mf <- SKAT.matrixfree(SKAT.example$Z, model=lm(y.c~1, data=SKAT.example))
Q<-skat2mf$Q()
all.equal(Q, 234803.786)
## correct value is 0.01874576
pQF(Q, skat2mf, neig=4, convolution.method="integration")

skat3mf <- SKAT.matrixfree(SKAT.example$Z, model=lm(y.c~X, data=SKAT.example))
Q<-skat3mf$Q()
all.equal(Q, 298041.542)
## correct value is 0.002877041
pQF(Q, skat3mf, neig=4, convolution.method="integration")

bigQF documentation built on Nov. 23, 2021, 5:06 p.m.