reverseDirection: reverseDirection

In SharonLutz/reverseDirection: Runs the MR Steiger approach on simulated data in the presence or absence of unmeasured confounding of the SNP-mediator-outcome relationship.

reverseDirection

Description

Runs the MR Steiger approach on simulated data in the presence or absence of pleiotropy, measurment error of the exposure, or unmeasured confounding of the exposure-outcome relationship.

Usage

reverseDirection(nSim = 1000, n = 100, MAF = c(rep(0.5, 10)), gamma0 = 0, gammaG = c(rep(0.2, 5), rep(0.1, 5)), varX = 1, measurementError = F, delta0 = 0, deltaX = 1, varME = 1, beta0 = 0, betaX = seq(from = 0, to = 1, length.out = 4), pleiotropy = F, betaG = c(rep(0.2, 5), rep(0.1, 5)), varY = 0.2, unmeasuredConfounding = F, meanU = 0, varU = 1, gammaU = 1, betaU = 1, sig.level = 0.05, SEED = 1, plot.pdf = T, plot.name = "reverseDirection")

Arguments

nSim	the number of simulations
n	the sample size
MAF	vector of the minor allele frequencies of the SNPs
gamma0	the intercept for the linear association of the SNPs G with the exposure X
gammaG	the vectore of slopes for the linear association of the SNPs G with the exposure X
varX	the variance of the exposure X which is generated from a normal distribution
measurementError	if measurementError=T then the exposure is generated with measurement error
delta0	the intercept for the linear association of the exposure with measurement error
deltaX	the slope for the linear association of the exposure with measurement error
varME	the variance of the exposure X with measurement error which is generated from a normal distribution
beta0	the intercept for the linear association of the exposure X with the outcome Y
betaX	the slope for the linear association of the exposure X with the outcome Y
pleiotropy	if pleiotropy=T then the outcome Y is generated with a direct effect of the SNP G on Y
betaG	the vector of slopes for the linear association of the SNPs G with the outcome Y
varY	the variance of the outcome Y which is generated from a normal distribution
unmeasuredConfounding	if unmeasuredConfounding=T then an unmeasured confounder U of the exposure X- outcome Y relationship is generated from a normal distribution with mean meanU and viarance varU
meanU	mean of the unmeasured confounder U
varU	variance of the unmeasured confounder U
gammaU	the effect of the unmeasured confounder U on the exposure X
betaU	the effect of the unmeasured confounder U on the outcome Y
sig.level	the significance level, default=0.05
SEED	the seed for the random number generator
plot.pdf	if plot.pdf=T then a plot of the percent of simulations where case 1-3 is accepted. If plot.pdf=F then no plot will be created.
plot.name	specifies the name of the plot that is created

Details

This function outputs the percent of simulations where case 1 (X->Y), case 2 (X<-Y), or case 3 (inconclusive) is detected by using the MR Steiger approach. The percent of simulations where the Steiger p-value and MR p-value is less than alpha is given (pSteiger and pMR, respectively). The correlation between the first SNP G1 and the exposure X (corG1X), correlation between the first SNP G1 and the outcome Y (corG1Y) and the correlation between the exposure X and the outcome Y (corXY).

Author(s)

Sharon Lutz

References

Hemani G, Tilling K, Davey Smith G (2017) Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLOS Genetics 13(11): e1007081.

Examples

reverseDirection(nSim=100)

SharonLutz/reverseDirection documentation built on Jan. 10, 2025, 4:40 a.m.