A learning-based framework for miRNA-disease association identification using neural networks

• DOI: 10.1093/bioinformatics/btz254
• Publication Date: 2019-04-06T11:21:32Z
• Authors: Jiajie Peng; Weiwei Hui; Qianqian Li; Bolin Chen; Jianye Hao; Qinghua Jiang; Xuequn Shang; Zhongyu Wei
• Keywords: Identification; Feature Learning; Similarity (geometry); Feature (linguistics); Association (psychology)

Abstract Motivation A microRNA (miRNA) is a type of non-coding RNA, which plays important roles in many biological processes. Lots studies have shown that miRNAs are implicated human diseases, indicating might be potential biomarkers for various types diseases. Therefore, it to reveal the relationships between and diseases/phenotypes. Results We propose novel learning-based framework, MDA-CNN, miRNA-disease association identification. The model first captures interaction features diseases based on three-layer network including disease similarity network, miRNA protein-protein network. Then, employs an auto-encoder identify essential feature combination each pair automatically. Finally, taking reduced representation as input, uses convolutional neural predict final label. evaluation results show proposed framework outperforms some state-of-the-art approaches large margin both tasks prediction miRNA-phenotype prediction. Availability implementation source code data available at https://github.com/Issingjessica/MDA-CNN. Supplementary information Bioinformatics online.