| Title: | Powerful Replicability Analysis of Genome-Wide Association Studies |
|---|---|
| Description: | A robust and powerful approach is developed for replicability analysis of two Genome-wide association studies (GWASs) accounting for the linkage disequilibrium (LD) among genetic variants. The LD structure in two GWASs is captured by a four-state hidden Markov model (HMM). The unknowns involved in the HMM are estimated by an efficient expectation-maximization (EM) algorithm in combination with a non-parametric estimation of functions. By incorporating information from adjacent locations via the HMM, this approach identifies the entire clusters of genotype-phenotype associated signals, improving the power of replicability analysis while effectively controlling the false discovery rate. |
| Authors: | Yan Li [aut, cre, cph], Haochen lei [aut], Xiaoquan Wen [aut], Hongyuan Cao [aut] |
| Maintainer: | Yan Li <[email protected]> |
| License: | GPL-3 |
| Version: | 1.0.1 |
| Built: | 2024-12-08 07:04:56 UTC |
| Source: | CRAN |
Estimate the rLIS values accounting for the linkage disequilibrium across two genome-wide association studies via the four-state hidden Markov model. Apply a step-up procedure to control the FDR of replicability null.
em_hmm(pa_in, pb_in, pi0a_in, pi0b_in)em_hmm(pa_in, pb_in, pi0a_in, pi0b_in)
pa_in |
A numeric vector of p-values from study 1. |
pb_in |
A numeric vector of p-values from study 2. |
pi0a_in |
An initial estimate of the null probability in study 1. |
pi0b_in |
An initial estimate of the null probability in study 2. |
rLIS |
The estimated rLIS for replicability null. |
fdr |
The adjusted values based on rLIS for FDR control. |
loglik |
The log-likelihood value with converged estimates of the unknowns. |
pi |
An estimate of the stationary probabilities of four states (0,0), (0,1), (1,0), (1,1). |
A |
An estimate of the 4-by-4 transition matrix. |
f1 |
A non-parametric estimate for the non-null probability density function in study 1. |
f2 |
A non-parametric estimate for the non-null probability density function in study 2. |
Replicability analysis across two genome-wide association studies accounting for the linkage disequilibrium structure.
ReAD(pa, pb)ReAD(pa, pb)
pa |
A numeric vector of p-values from study 1. |
pb |
A numeric vector of p-values from study 2. |
A list:
rLIS |
The estimated rLIS for replicability null. |
fdr |
The adjusted values based on rLIS for FDR control. |
loglik |
The log-likelihood value with converged estimates of the unknowns. |
pi |
An estimate of the stationary probabilities of four states (0,0), (0,1), (1,0), (1,1). |
A |
An estimate of the 4-by-4 transition matrix. |
f1 |
A non-parametric estimate for the non-null probability density function in study 1. |
f2 |
A non-parametric estimate for the non-null probability density function in study 2. |
# Simulate p-values in two studies locally dependent via a four-state hidden Markov model data <- SimuData(J = 10000) p1 = data$pa; p2 = data$pb; theta1 = data$theta1; theta2 = data$theta2 # Run ReAD to identify replicable signals res.read = ReAD(p1, p2) sig.idx = which(res.read$fdr <= 0.05)# Simulate p-values in two studies locally dependent via a four-state hidden Markov model data <- SimuData(J = 10000) p1 = data$pa; p2 = data$pb; theta1 = data$theta1; theta2 = data$theta2 # Run ReAD to identify replicable signals res.read = ReAD(p1, p2) sig.idx = which(res.read$fdr <= 0.05)
Simulate two sequences of p-values by accounting for the local dependence structure via a hidden Markov model.
SimuData( J = 10000, pi = c(0.25, 0.25, 0.25, 0.25), A = 0.6 * diag(4) + 0.1, muA = 2, muB = 2, sdA = 1, sdB = 1 )SimuData( J = 10000, pi = c(0.25, 0.25, 0.25, 0.25), A = 0.6 * diag(4) + 0.1, muA = 2, muB = 2, sdA = 1, sdB = 1 )
J |
The number of features to be tested in two studies. |
pi |
The stationary probabilities of four hidden joint states. |
A |
The 4-by-4 transition matrix. |
muA |
Mean of the normal distribution generating the p-value in study 1. |
muB |
Mean of the normal distribution generating the p-value in study 2. |
sdA |
The standard deviation of the normal distribution generating the p-value in study 1. |
sdB |
The standard deviation of the normal distribution generating the p-value in study 2. |
A list:
pa |
A numeric vector of p-values from study 1. |
pb |
A numeric vector of p-values from study 2. |
theta1 |
The true states of features in study 1. |
theta2 |
The true states of features in study 2. |