seqMeta/MetaSKAT

Performing a survival meta-analysis of rare variants from sequencing data using seqMeta/MetaSKAT

Instalation SKAT/ MetaSKAT/seqMeta

• Use R Version < 4.0.2 if you got a instalation "had non-zero exit status" error & and fix C++/g++/ rtools.

Required packages:

setwd("D:/folder")
install_github("brentp/skatMeta")

library(CompQuadForm)
library(twostageGWASsurvival)
library(seqMeta)
library(MASS)
library(writexl)
library(Matrix)

Required data files:

1- SNP information (SNPInfo): contains SNP & gene names (establishing a connection between the genetic loci and their corresponding genes)


2- Demographic & clinical data file (data): Includes age, sex, Bp, survival time, outcome variable , ...for each individual.

3- Z matrix (Genotype Matrix): This matrix represents the genotype data for the genetic variants. The rows of this matrix correspond to individual samples, while the columns indicate the SNP names (genetic loci). The entries of the matrix represent the alleles present at each SNP for each individual. It is generally a binary matrix, where 0 denotes one allele and 1 denotes the other allele.


To create Z we need allele data set contains individuals ID and position:

           allel_binary_data.txt

Pos	Ref	Alter	sample 1	sample 2	sample 3
45	A	G	1	1	0
158	TG	G	0	1	0
86	T	C	0	0	0

genomatrix = read.table("allel_binary_data.txt",header=TRUE)
snpInfo = read.csv('SNPInfo.csv',header=TRUE)
names(genomatrix)

I = match(snpInfo$POS, genomatrix$POS)
genomatrix = genomatrix[I,]
genomatrix = t(genomatrix[,-c(1:3)])
colnames(genomatrix) = snpInfo$SNP

Runing survival model

cox_model<-prepCox(genomatrix, Surv(time,binary_outcome) ~ Age +  Sex +  intervention, 
             SNPInfo = SNPInfo, data = my_data, verbose = FALSE)

output_cox <- skatMeta(cox_model, SNPInfo = SNPInfo)
head(output_cox)

SKAT Optimal Test

SKAT O Test introduced to test weighted averages of SKAT & burden tests

cox_model1<- prepScores(genomatrix_a, outcome ~ sex+age+ intervention, 
             SNPInfo = SNPInfo, data = my_data, verbose = FALSE)

cox_model2<-prepScores(genomatrix_b, outcome ~ sex+age+ intervention, 
             SNPInfo = SNPInfo, data = my_data_b, verbose = FALSE)


Out_combined <- skatOMeta(cox_model1,cox_model2, SNPInfo=SNPInfo, method="int")

head(Out_combined)

##Survival data

#survival model 1
Cox_model1<-prepCox(genomatrixS, Surv(time,rel) ~ sex+age+ intervention,
                    SNPInfo = SNPInfo, data = my_data, verbose = FALSE)

out.Cox_model1 <- skatOMeta(Cox_model1, SNPInfo = SNPInfo)
head(out.Cox_model1)

#survival model 2
Cox_model2<-prepCox(genomatrix_b, Surv(time,rel) ~   sex+age+ intervention, 
             SNPInfo = SNPInfo, data = my_data_b, verbose = FALSE)

out_Cox_model2 <- skatMeta(Cox_model2, SNPInfo = SNPInfo)
head(out.Cox_model2)


overall_Cox <- skatOMeta(Cox_model1, Cox_model2, SNPInfo = SNPInfo)
head(overall_Cox)

Results from the SKAT test

gene	p	pmin	rho	cmaf	nmiss	nsnps	errflag
Name of Gene	P-value: is <0.05?	Min P-value	𝜌 (0, 1)	Cumulative minor allele frequency	n of missing SNPs	n SNPs in the gene	Inaccurate p-values

Ref.

seqMeta

seqMeta: an R Package for meta-analyzing region-based tests of rare DNA variants

Meta-MultiSKAT: Multiple phenotype meta-analysis for region-based association test