|
Abstract:
|
Integration of multiple genetic sources for copy number variation detection is a powerful approach to improve the identification of variants associated with complex traits. Although it has shown that the widely used change point model based methods can increase statistical power to identify variants, it remains challenging to effectively integrate an important type of information, the allele intensity ratio signal, B allele frequency (BAF), into these methods. We previously developed modSaRa, a normal mean based model on a screening and ranking algorithm for copy number variation identification with high computational efficiency. Here we proposed a novel improvement of modSaRa by incorporating BAF into the change point model to boost statistical power for the identification of variants. Furthermore, a semi-supervision likelihood based segmentation was developed for accurate copy number segmentation. Simulation studies illustrate its power to improve the identification of generic variants. The application of the new method to a whole genome melanoma dataset facilitates understanding of the possible roles of germline copy number variants in complex diseases.
|
