-
Notifications
You must be signed in to change notification settings - Fork 0
Home
Diptavo Dutta edited this page Jan 22, 2019
·
4 revisions
This R-package tests for rare-variant association tests with multiple phenotypes by meta-analysis across multiple studies. The functions aggregate variant-phenotype score statistic in a particular region/gene and computes corresponding asymptotic p-values.
In this vignette we display an elementary workflow to obtain MetaMultiSKAT test results (omnibus and with pre-specified kernel)
The first step is to save the summary-statistics for each study.
library(MetaMultiSKAT)
data(MultiSKAT.example.data)
attach(MultiSKAT.example.data)
### Create null model
obj.null <- MultiSKAT_NULL(Phenotypes,Cov)
### Perform a MultiSKAT Phenotype-Kernel: PhC; Genotype-Kernel: SKAT
out1 <- MultiSKAT(obj.null,Genotypes,Sigma_p = cov(Phenotypes),verbose = FALSE)
S1 <- MetaMultiSKAT:::Extract.Test.Info(obj.null,Genotypes)
MetaMultiSKAT:::save.Meta.Files(S1,"S1",study = "1")
Once the individual summary statistic files are saved and available the next step is to read in the files. There are a 3 such data files available with the package as an example.
s1 <- system.file("RDS_data","S1_st.1.RDS", package = "MetaMultiSKAT",mustWork = T)
s2 <- system.file("RDS_data","S2_st.2.RDS", package = "MetaMultiSKAT",mustWork = T)
s3 <- system.file("RDS_data","S3_st.3.RDS", package = "MetaMultiSKAT",mustWork = T)
study.list <- MetaMultiSKAT:::read.Meta.Files(c(s1,s2,s3))
Subsequently the standard MetaMultiSKAT tests can be done as follows
Meta.Het(study.list,verbose = F,resample = 100)$p.value
Meta.Hom(study.list,verbose = F,resample = 100)$p.value
Meta.Com(study.list,verbose = F,resample = 100)$p.value