Detection of copy number variation from next generation sequencing data with total variation penalized least square optimization

The detection of copy number variation is important to understand complex diseases such as autism, schizophrenia, cancer, etc. In this paper we propose a method to detect copy number variation from next generation sequencing data. Compared with conventional methods to detect copy number variation like array comparative genomic hybridization (aCGH), the next generation sequencing data provide higher resolution of genomic variations. There are a lot of methods to detect copy number variation from next sequencing data, and most of them are based on statistical hypothesis testing. In this paper, we consider this problem from an optimization point of view. The proposed method is based on optimizing a total variation penalized least square criterion, which involves ℓ-1 norm. Inspired by the analytical study of a statics system, we propose an iterative algorithm to find the optimal solution of this optimization problem. The comparative study with other existing methods on simulated data demonstrates that our method can detect relatively small copy number variants (low copy number and small single copy length) with low false positive rate.