prsMultitest: Average Over Multiple PRS Simulations

In andreyshabalin/simPRS: Fast Simulation of Polygenic Risk Scores

Description

Perform a number of PRS simulations with gwasFast and prsInf and average over the results. Supports parallelization on multicore CPUs.

Usage

prsMultitest(signal, N, Nsim, nthreads = 0, minpv = 1e-20)

Arguments

signal	Vector of effect sizes for all SNPs. Can be generated with genSignal.
N	GWAS sample size.
Nsim	Number of simulations to perform.
nthreads	Number of parallel threads to use. Zero (default) set it to the number of CPU cores. Setting nthreads = 1 disables multithreading.
minpv	The simulated PRS performance is recorded on the interval from minpv to 1.

Details

The function uses parallel package for multithreaded calculations.

Value

The function returns the vector of p-value thresholds and corresponding average asymptotic correlations of the phenotype with PRS (r).

Author(s)

Andrey A Shabalin [e-mail]

Examples

NTotalSNPs = 10000
NSignalSnps = 100
heritability = 0.2
signalDistr = "Same"
Ntrain = 10000
Ntest = 3000

signal = genSignal(
            NSignalSnps = NSignalSnps,
            NTotalSNPs = NTotalSNPs,
            heritability = heritability,
            signalDistr = signalDistr)
            
prsA = prsMultitest(signal = signal, N = Ntrain, Nsim = 30, nthreads = 1)

rci = rConfInt(r = prsA$r, N = Ntest)

prsPlot(pv = prsA$pv, r = prsA$r, rci)

andreyshabalin/simPRS documentation built on May 21, 2019, 2 p.m.