A practical case-control association test for detecting a susceptibility allele at a copy number variation locus

The association between a copy number variant (CNV) and susceptibility to disease has drawn much attention. In this study, a case-control association test for a CNV locus with multiple alleles is proposed for detecting a single CNV allele associated with a disease. In the association test, CNV allele frequencies are estimated for cases and controls separately using an expectation-maximization (EM) algorithm, and the chi(2) values are calculated for each CNV allele to compare the estimated frequency between them. A permutation procedure is used to obtain an empirical P-value for each CNV allele and for controlling a global type I error rate. The statistical power of the present association test was evaluated by a computer simulation analysis with several parameter settings. The results revealed that the statistical power was markedly different among CNV alleles with different copy numbers, and a higher power could be achieved for a susceptible allele with the lowest or highest copy number in comparison with those with intermediate copy numbers.