稀有变量关联分析中的泊松近似变阈值方法

Poisson approximation variable threshold method in rare variants association study

稀有基因变量与疾病性状的关联性研究是全基因组关联分析的补充.然而,传统的关联分析检验方法不适用于稀有变量的检验问题,因此众多的针对稀有变量的新检验方法得以被提出和运用.变阈值方法是其中的一种方法,该方法相对于传统的关联分析检验方法在功效上有较大提高.这里从稀有事件的角度看待稀有变量突变等位基因的发生,以泊松近似分布代替二项分布,提出新的检验统计量,并在变阈值方法的基础上给出一个新的检验方法,即泊松近似变阈值方法.模拟试验结果表明该方法比变阈值方法有一致更高的检验功效,筛选变量的效率也得到了提高.

The association study of rare genetic variants and disease traits as a supplement of genome-wide association study（GWAS）has achieved remarkable development and application.But the traditional association testing methods are not suitable for the rare variants association problem.Many new testing methods designed specifically for the rare variants have been proposed and applied.The variable-threshold（VT）method is one of them,which has achieved more power than the traditional association test methods.Here the occurrence of the mutation of rare variants was treated as a rare event.Poisson distribution was taken instead of binomial distribution and a new test statistic was proposed.Base on the VT method,a new test method was proposed,namely,the Poisson approximation variable threshold method（PAVT）.Simulation results show that the new method has more power than VT method uniformly,and variant selection is more efficient.