GEM_Gmodel: GEM_Gmodel Analysis

In fastGEM/GEM: GEM: fast association study for the interplay of Gene, Environment and Methylation

Description

GEM_Gmodel creates a methQTL genome-wide map.

Usage

GEM_Gmodel(snp_file_name, covariate_file_name, methylation_file_name, Gmodel_pv,
  output_file_name)

Arguments

snp_file_name	Text file with rows representing genotype encoded as 1,2,3 or any three distinct values for major allele homozygote (AA), heterozygote (AB) and minor allele homozygote (BB) and columns representing samples, such as the example data file "snp.txt".
covariate_file_name	Text file with rows representing covariate factors and columns representing samples, such as the example data file "cov.txt".
methylation_file_name	Text file with rows representing methylation profiles for CpGs, and columns representing samples, such as the example data file "methylation.txt".
Gmodel_pv	The pvalue cut off. Associations with significances at Gmodel_pv level or below are saved to output_file_name, with corresponding estimate of effect size (slope coefficient), test statistics and p-value. Default value is 5.0E-08.
output_file_name	The result file with each row presenting a CpG and its association with SNP, which contains CpGID, SNPID, estimate of effect size (slope coefficient), test statistics, pvalue and FDR at each column.

Details

GEM_Gmodel creates a methQTL genome-wide map by performing matrix based iterative correlation and memory-efficient data analysis instead of millions of linear regressions (N = number_of_CpGs x number_of_SNPs) between methylation and genotyping. Polymorphisms close to CpGs in the same chromosome (cis-) or different chromosome (trans-) often form methylation quantitative trait loci (methQTLs) with CpGs. In GEM_Gmodel, MethQTLs can be discovered by correlating single nucleotide polymorphism (SNP) data with CpG methylation from the same samples, by linear regression lm (M ~ G + covt), where M is a matrix with methylation data, G is a matrix with genotype data and covt is a matrix with covariates, and all read from the formatted text data file. The methylation data are the measurements for CpG probes, for example, 450,000 CpGs from Illumina Infinium HumanMethylation450 Array. The genotype data are encoded as 1,2,3 or any three distinct values for major allele homozygote (AA), heterozygote (AB) and minor allele homozygote (BB). The linear regression is adjusted by covariates read from covariate data file. The output of GEM_Gmodel is a list of CpG-SNP pairs, where the SNP is the best fit to explain the particular CpG. The significant association between CpG-SNP pair suggests the methylation driven by genotyping variants, which is so called methylation quantitative trait loci (methQTL).

Value

save results automatically

Examples

DATADIR = system.file('extdata',package='GEM')
RESULTDIR = getwd()
snp_file = paste(DATADIR, "snp.txt", sep = .Platform$file.sep)
covariate_file = paste(DATADIR, "cov.txt", sep = .Platform$file.sep)
methylation_file = paste(DATADIR, "methylation.txt", sep = .Platform$file.sep)
Gmodel_pv = 1e-04
output_file = paste(RESULTDIR, "Result_Gmodel.txt", sep = .Platform$file.sep)
GEM_Gmodel(snp_file, covariate_file, methylation_file, Gmodel_pv, output_file)

fastGEM/GEM documentation built on Aug. 5, 2021, 7:33 p.m.