Data fusion-based algorithm for predicting miRNA-Disease associations

Abstract

Technological progress and the development of laboratory techniques and bioinformatics tools have led to the availability of ever-increasing amounts of biological data including genomic, proteomic, and transcriptomic sequences and related information. These data have helped in understanding some of the complicated life process from a systematic level. Many diseases are generated by abnormalities in multiple regulating processes. In this study, we constructed a novel miRNA-gene-disease fusion (MGDF) algorithm by integrating three genome-wide networks, namely microRNA (miRNA), gene function, and disease similarity networks. The data fusion method was applied to construct a miRNA-gene-disease association network model from these networks to explore miRNA-disease associations mediated by genes with similar functions. mmiRNAs bind to their target genes and regulate their expression, so the miRNA-gene and gene-disease regulatory relationships were included in the network model to more accurately predict miRNA-disease associations. The proposed MGDF was used to predict miRNA-cancer associations and the results show that most of the predicted associations had evidence in existing databases.

Keywords: Disease; Network fusion; Random walk; miRNA.