Predicting lincRNA-Disease Association in Heterogeneous Networks Using Co-regularized Non-negative Matrix Factorization

• Main Authors: Yong Lin, Xiaoke Ma
• Format: Article
• Language: English
• Published: Frontiers Media S.A. 2021-01-01
• Series: Frontiers in Genetics
• Subjects: disease-lincRNA association; non-negative matrix factorization; heterogeneous network; regularization; network analysis
• Online Access: https://www.frontiersin.org/articles/10.3389/fgene.2020.622234/full

Long intergenic non-coding ribonucleic acids (lincRNAs) are critical regulators for many complex diseases, and identification of disease-lincRNA association is both costly and time-consuming. Therefore, it is necessary to design computational approaches to predict the disease-lincRNA associations that shed light on the mechanisms of diseases. In this study, we develop a co-regularized non-negative matrix factorization (aka Cr-NMF) to identify potential disease-lincRNA associations by integrating the gene expression of lincRNAs, genetic interaction network for mRNA genes, gene-lincRNA associations, and disease-gene associations. The Cr-NMF algorithm factorizes the disease-lincRNA associations, while the other associations/interactions are integrated using regularization. Furthermore, the regularization does not only preserve the topological structure of the lincRNA co-expression network, but also maintains the links “lincRNA → gene → disease.” Experimental results demonstrate that the proposed algorithm outperforms state-of-the-art methods in terms of accuracy on predicting the disease-lincRNA associations. The model and algorithm provide an effective way to explore disease-lncRNA associations.