EMSATS: PCA based adapitve SNP-set association test of multiple...

In baolinwu/MSKAT: Various multi-trait variant-set association test methods

Description

We use the LD score regression (Bulik-Sullivan et al.) to accurately estimate the trait correlation, which is then used to construct the multi-trait association tests of multiple variants in a gene or pathway: PC based test (ET), variance components test (VT), and their adaptive test (AT). ET works well for mediated pleiotropy (see ref).

Usage

EMSATS(Z, Sig, R, rho = c(-10, -5, 2, -1, 0:5/5))

Arguments

Z	M by K matrix of summary Z-statistics for M variants across K traits
Sig	estimated trait correlation matrix (K by K)
R	variant LD correlation matrix (M by M)
rho	weight assigned to the ET

Value

p.value

test p-values for: AT, VC, ET

pval

vector of all p-values

rho.est

the optimal rho weight

References

Guo,B. Liu,N. and Wu,B. (2018) Integrate multiple traits and multiple variants to detect novel disease-gene association using GWAS summary data. tech rep.

Examples

K = 4; M = 20
R = cor(matrix(rnorm(500*M),500,M)*sqrt(0.8)+rnorm(500)*sqrt(0.2))
Y = matrix(rnorm(100*K), 100,K)*sqrt(0.8)+rnorm(100)*sqrt(0.2)
Y[,1] = -Y[,1]; Sig = cor(Y)
Z = matrix(rnorm(K*M),M,K)
js = sample(1:M, size=round(M*0.4))
Z[js,] = Z[js,] + 1.5
EMSATS(Z,Sig,R)
Z = matrix(rnorm(K*M),M,K)
ij = cbind(sample(1:M, size=20, rep=TRUE), sample(1:K, size=20, rep=TRUE))
Z[ij] = Z[ij] + rnorm(20)*1.5
EMSATS(Z,Sig,R)

baolinwu/MSKAT documentation built on May 28, 2019, 6:37 p.m.