dmvc {DMtest}

Perform DMC, DVC, DMVC, and DMVC+ tests for genome-wide CpGs in methylation arrays.

Description

This function implements an algorithm for computing various tests of mean and variance differences, including the DMVC+ test that specifically addresses the hypermethylation and hypervariability for cancer-specific CpGs

Usage

dmvc(beta = beta, covariate = covariate, npermut=100,permut.seed=100,
corenumber=1)

Arguments

beta

Methylation beta value matrix, row for CpGs, column for samples. The matrix has sample name as the column names, and CpG names as the row names.

covariate

covariate matrix, a data frame including all covariates in the regression model, whose row represents for samples, column represents different covariates. The matrix has sample names as the row names. The matrix must include a "group" column, which is a binary indicator (0 for normal and 1 for tumor) to define two groups of samples to be compared.

npermut

The number of permutations for computing the correlation that is needed for the joint tests

permut.seed

The random seed used by permutation for joint tests

corenumber

The number of cores to be used for joint tests; if corenumber>1, a parallel computing version will be used to speed up the computation

Value

A data frame with the following columns.

Mean_normal

Mean of beta values for normal samples.

Mean_tumor

Mean of beta values for tumor samples.

Mean_all

Mean of beta values for all samples.

SD_normal

Standard deviation of beta values for normal samples.

SD_tumor

Standard deviation of beta values for tumor samples.

SD_all

Standard deviation of beta values for all samples.

DMCP

p-value from DMC test.

DVCP

p-value from DVC test.

Joint1P

Joint test for DMVC+ (test for hypermethylation and increased variance in cancer samples).

Joint2P

Joint test for DMVC (test for differential methylation in both direction and increased variance in cancer samples).

LRT1

Likelihood ratio test statistics for joint test1.

LRT2

Likelihood ratio test statistics for joint test2.

pho

Correlation value computed by permutations.

References

Dai, J, Wang, X, Chen, H and others. (2021). Incorporating increased variability in discovering cancer methylation markers, Biostatistics, submitted.

Examples

data(beta)
data(covariate)
out=dmvc(beta=beta,covariate=covariate)