CCGA-package: Case-control genetic association analysis with adjustment of...

In zhanghfd/CCGA: Case-control genetic association analysis incorporating non-confounding covariates

Description

In CCGA, the power of disease-SNP association analysis in case-control studies can be potentially improved through efficiently adjustmenting non-confounding covariates, compared with the standard logistic regression method with or without covariate adjustment.

Let Y, S, G, and X denote the case-control status, stratum variable, SNP genotype coded by 0, 1, or 2, and covariate(s), respectively. The model for relating the response Y and (S,G,X) is

g(pr(Y=1)) = alpha + beta_S S + beta_G G + beta_X X,

where g(.) is a given function, which can be the logit function or the probit function. The disease prevalence is required for each stratum.

The disease prevalence(s), Hardy-Weinberg equilibrium, and independence between non-confounding covariate(s) and SNP genotype are efficiently incorporated in the retrospective likelihood function, and the nonparametric distribution of the covariate(s) is profiled out through the application of Lagrange's multipler method. The multipliers can be directly estimated using the available data, which yields the so called profile maximum likelihood estimates of the regression parameter (pMLE). Alternatively, the data-dependent multiplier(s) can be replaced with the limiting value(s) to yield a modified profile likeihood function, which results in modified profile maximum likelihood estimates (mpMLE). Theoretically, mpMLE and pMLE are asymptotically equivalent in terms of estimation efficiency, but mpMLE could be computationally much simplier and faster.

This package includes two main functions (i.e., SingleSNP and MultipleSNP) and a simulated dataset for illustration. In SingleSNP and MultipleSNP, two candidate link functions (i.e., the logit function and probit function) can be used, and both mpMLE and pMLE can be implemented. In the function MultipleSNP, multiple CPU cores can be used to speed up the analysis with UNIX-like OS.

Details

Package:	CCGA
Type:	Package
Version:	1.0
Date:	2016-10-27
License:	Artistic License 2.0

Author(s)

Hong Zhang

Maintainer: Hong Zhang <[e-mail]>

References

Zhang H, Chatterjee N, Rader D, Chen J. (2016) Adjustment of Non-confounding Covariates in Case-control Genetic Association Studies. Annals of Applied Statistics (revised).

zhanghfd/CCGA documentation built on May 4, 2019, 10:16 p.m.