Chromatin accessibility prediction via a hybrid deep convolutional neural network

• DOI: 10.1093/bioinformatics/btx679
• Publication Date: 2017-10-20T11:22:27Z
• Authors: Qiao Liu; Fei Xia; Qijin Yin; Rui Jiang
• Keywords: DECIPHER; Identification

A majority of known genetic variants associated with human-inherited diseases lie in non-coding regions that lack adequate interpretation, making it indispensable to systematically discover functional sites at the whole genome level and precisely decipher their implications a comprehensive manner. Although computational approaches have been complementing high-throughput biological experiments towards annotation human genome, still remains big challenge accurately annotate regulatory elements context specific cell type via automatic learning DNA sequence code from large-scale sequencing data. Indeed, development an accurate interpretable model learn signature further enable identification causative has become essential both genomic studies.We proposed Deopen, hybrid framework mainly based on deep convolutional neural network, automatically sequences predict chromatin accessibility. In series comparison existing methods, we show superior performance our not only classification accessible against background sampled random, but also regression DNase-seq signals. Besides, visualize kernels match identified signatures motifs. We finally demonstrate sensitivity finding noncoding analysis breast cancer dataset. expect see wide applications Deopen either public or in-house accessibility data diseases.Deopen is freely available https://github.com/kimmo1019/[e-mail] are Bioinformatics online.