A geometric deep learning framework for drug repositioning over heterogeneous information networks

Abstract

The effectiveness of computational drug repositioning techniques has been further improved due to the development of artificial intelligence technology. However, most of the existing approaches fall short of taking into account the non-Euclidean nature of biomedical data. To overcome this problem, we propose a geometric deep learning (GDL) framework, namely DDAGDL, to predict drug-disease associations (DDAs) on heterogeneous information networks (HINs). DDAGDL can take advantage of complicated biological information to learn the feature representations of drugs and diseases by ingeniously projecting drugs and diseases including geometric prior knowledge of network structure in a non-Euclidean domain onto a latent feature space. Experimental results show that DDAGDL is able to identify high-quality candidates for Alzheimers disease (AD) and Breast neoplasms (BN) that have already been reported by previously published studies, and some of them are not even identified by comparing models.