Genetic Association Test Of Multi-trait Genetic Data

With the rapid development of gene sequencing technology,people can explore the cause of the complex diseases at the molecular level.Because most of the complex diseases are determined by genes,so researchers can get precise diagnosis by testing for associations between diseases and genetic loci.Research has shown that besides common varians,some rare variants also explain part of the phenotypic variance for human diseases.Due to the importance of rare variants,some statistical methods need to be proposed to test for the association between the rare variants and disease.In addition,more and more rare genetic variants are being detected in the human genome,which also provides the possibility to research the rare variants.The complex diseases are usually recorded as a variety of forms:such as binary,ordinal,quantitative and/or any mixture of them.The complex diseases are often regarded as the association between multivariate phenotypes and multiple loci;therefore,we can't only look at the association between the single trait with multiple loci and the joint information of multiple traits need to be considered simultaneously and sufficiently.Since family members have more chances to share the same causal mutation,so the genetic heterogeneity is relatively small in the family.It will be a good strategy to test the association between the rare variants and multiple traits based on family design.In this article,under the hypothesis of common diseases of rare variants,we construct a new nonparametric statistics U based on the generalized Kendall's ?theory,and futher consider the asymptotic distribution of the proposed association test statistic.The new statistical method is used to study the association between the rare variants and multiple traits,and it broadens the way of thinking on the mapping problem of disease genes.By comparing with the existing mature classic methods,it is concluded that the proposed method in this article has enforceability and validity.This paper describes five methods based on individuals:CAST,SUM,SKAT,SKAT-0 and NM-RV,and emphasizes on the method of NM-RV-FAM which is based on family data of trios.Through the application of simulated data and real data of COL6A3,we find that the type I error rate of six kinds of methods are all within the confidence interval.For the power,the NM-RV-FAM method has high powers,especially for the high risk factors,and the powers of cases of two ordinal traits and mixed traits are higher than those of two binary.In our opinion,the proposed method in this article is powerful and robust in testing for associations between rare variants and multiple traits.