유전변이 분석을 위한 비모수 가설 검정 방법

We propose a method of the hypothesis test for genetic association analysis. Conventional methods represent the genotype values as the number of different allele or its variants, and perform a usual parametric test, such as t-test. However, the genotype is an ordinal quantity which do not have a magnitude. Thus, the conventional method is not robust under how the genotype is encoded. This shortcoming becomes prominent if one uses the principle component analysis in order to correct for the population structure. To overcome this shortcoming, in the presentation, we suggest a Kendall statistics which evaluates the degree of similarity between two sets of ranks. Kendall statistics is more or less independent of the encoding and free from the assigning numerical value of allele. In addition, the Kendall statistics does not require the normality of the phenotype values. This is another advantage of the usage of Kendall statistics, because phenotype values obtained from the dominant effect are not, in general, distributed as a Gaussian distribution. Experimental results support the advantage of the proposed method over the conventional models.