Prediction of disease-associated circRNAs via circRNA–disease pair graph and weighted nuclear norm minimization

Abstract

Circular RNA (circRNA) is a special single-stranded, non-coding RNA molecule. A variety of circRNAs are widely distributed in organisms and can regulate gene expression by adsorbing miRNAs and proteins. Therefore, abnormal expression of circRNAs can reflect a variety of diseases. In this study, from a computational perspective, potential circRNA–disease relationships are predicted by a novel method named PDC-PGWNNM. A pair graph of circRNA–disease is designed using the interactions of circRNAs with miRNAs and regulatory relationship of miRNAs in diseases. After extracting a reliable negative set, the original observation set is expanded, and a weighted nuclear norm minimization model is used to predict the potential circRNA–disease associations. Experimental results show that PDC-PGWNNM has better performance than existing methods. At the same time, the rationality of the pair graph is explored. The regulatory information of miRNAs and the expression profiles of circRNAs are used to analyse the predicted disease–circRNAs. Special cases are discussed for glioma, colorectal cancer and coronary heart disease.