Association Analysis Method For Rare Variants Of Ordinal Responses

Genome-wide association studies have successfully identified genetic variants asso-ciated with complex diseases and these variants are mainly common variants. But theseidentified variants can only explain a small proportion of inheritable phenotypic variance.Studies show that complex diseases are influenced by multiple rare variants. The advancesin next-generation sequencing technologies offer us a lot of data about rare variants andmake more and more researchers begin studying the association between rare variants andcomplex diseases. As the frequency of rare variants is very low, traditional associationmethods are not applicative. Though there are some association tests about rare variants,they all select singly dichotomous or continuous traits as responses. We often simulta-neously regard multiple traits associated with diseases in our life, especially when westudy mental behavior disorders. Each trait is ordinal data about the severity of disease.We study the association for the rare variants of multiply ordinal responses in this paper.We establish cumulative logistic regression model for multiply ordinal responses and usethe generalized estimating equations for estimating parameters. Testing the parametersthrough the hypothesis conducts the association study. We compare with other existingmethods through extensive simulated data and the results show that our approach is morepowerful.