Deep learning frameworks for protein–protein interaction prediction
Xiaotian Hu, Cong Feng, Tianyi Ling, Ming Chen
Abstract
Protein-protein interactions (PPIs) play key roles in a broad range of biological processes. The disorder of PPIs often causes various physical and mental diseases, which makes PPIs become the focus of the research on disease mechanism and clinical treatment. Since a large number of PPIs have been identified by in vivo and in vitro experimental techniques, the increasing scale of PPI data with the inherent complexity of interacting mechanisms has encouraged a growing use of computational methods to predict PPIs. Until recently, deep learning plays an increasingly important role in the machine learning field due to its remarkable non-linear transformation ability. In this article, we aim to present readers with a comprehensive introduction of deep learning in PPI prediction, including the diverse learning architectures, benchmarks and extended applications.