Copy number variation (CNV) is a kind of genetic variation with abnormal fragment copies. Current algorithms for CNV association study for whole genome sequencing are restricted to specific size or common/rare CNVs. In this study, we construct a Bayesian procedure to detect disease associated CNVs. First, the absolute copy number of each window is estimated from sequencing read depths for every samples. And then the absolute copy numbers from case and control are compared to select candidate disease associated windows. Finally, the information from neighboring windows are combined to identify the disease associated copy number regions.