MetaCor-package: An R package to perform genetic association study with...

In tamartsi/MetaCor: Performs Stratified SNP-trait analysis and combines the results using the MetaCor method

Description

Analyzed stratified a stratigied genetic association study with correlated individuals between strata

Code used to perform stratified SNP-trait association study when individuals are correlated (and possibly overlapping) between the strata. Effect estimates from the individual strata are meta-analyzed using MetaCor. Tamar Sofer Tamar Sofer with contributions from Matt Conomos

Maintainer: Tamar Sofer Tamar Sofer <[e-mail]>

#### load the data - note: all data in the package is simulated and is very simple. ### load a data frame with genotype data data(G) G[1:3,] ## note: sex_stratInd are indicators for selection of a maximum set of participants if we want the two sex strata independent.

## load a data frame with covariates info data(W) head(W)

## load the outcome data(Y) head(Y)

## load the simplified kinship matrix data(K) K[1:5,1:5] ## uncorrelated individuals since they are from different families K[which(W$ID_fam == 1),which(W$ID_fam == 1)] ## these individuals are from the same family and have kinship coefficient 0.5.

## Prepare to fitting a stratified model with strata of sex - create a list of participants IDs in each of the strata: ## check: all(W$ID == G$ID) #TRUE IDs.list.gender <- list(male = W$ID[which(W$sex == 1)], female = W$ID[which(W$sex == 0)]) IDs.list.gender.stratInd <- list(male = W$ID[which(W$sex == 1 & W$sex_stratInd == 1)], female = W$ID[which(W$sex == 0 & W$sex_stratInd == 1)])

### fit a stratified model for males and females separately, and combine the results using MetaCor using a single function that does it all: strat.model <- stratLMMTest(Y, as.matrix(W[,c("intercept", "bmi", "bmi2", "age")]), as.matrix(G[,grep("SNP", colnames(G))]), covMatList = list(kinship = K), IDsList = IDs.list.gender, verbose = T, testType = "MetaGLS")

### When according to the kinship matrix, the two strata are indpendent... (and also trying to use MetaCor1, names here "MetaCor.fixef") strat.model <- stratLMMTest(Y, as.matrix(W[,c("intercept", "bmi", "bmi2", "age")]), as.matrix(G[,grep("SNP", colnames(G))]), covMatList = list(kinship = K), IDsList = IDs.list.gender.stratInd, verbose = T, testType = "MetaCor.fixef")

## if the list of IDs provides only a single stratum, a pooled model is fit: pooled.model <- stratLMMTest(Y, as.matrix(W[,c("intercept", "bmi", "bmi2", "age", "sex")]), as.matrix(G[,grep("SNP", colnames(G))]), covMatList = list(kinship = K), IDsList = list(pooled = rownames(W)), verbose = T, testType = "MetaCor.fixef")

tamartsi/MetaCor documentation built on May 31, 2019, 2:56 a.m.