Parallel CNV detection algorithm based on cloud computing

Recently, the cost of whole-genome sequencing has decreased dramatically due to the development of next-generation sequencing technology, and a huge amount of sequencing data has been generated and released by research laboratories worldwide. However, it is difficult to develop mature genome analysis software and high-performance computing resources which are available to assay genome data in real time. In this paper, we propose a parallel and robust CNV detection algorithm to run on a cloud computing environment. The proposed method, which we call CNV shape was developed using a shape-based CNV detection algorithm with Map/Reduce framework. This method finds regions above a certain length with continuously increased or decreased read coverage generated by mapping sequencing data onto the human reference genome. In order to maintain load balancing of each node in the cloud computing environment, we use a partitioning method. Also, we demonstrate the efficiency of the proposed method for CNV detection using publicly available sequencing data.