Protein–protein interface prediction based on hexagon structure similarity

Research output: Journal Publications and ReviewsRGC 21 - Publication in refereed journalpeer-review

14 Scopus Citations
View graph of relations



Original languageEnglish
Pages (from-to)83-88
Journal / PublicationComputational Biology and Chemistry
Online published12 Feb 2016
Publication statusPublished - Aug 2016


Studies on protein–protein interaction are important in proteome research. How to build more effective models based on sequence information, structure information and physicochemical characteristics, is the key technology in protein–protein interface prediction. In this paper, we study the protein–protein interface prediction problem. We propose a novel method for identifying residues on interfaces from an input protein with both sequence and 3D structure information, based on hexagon structure similarity. Experiments show that our method achieves better results than some state-of-the-art methods for identifying protein–protein interface. Comparing to existing methods, our approach improves F-measure value by at least 0.03. On a common dataset consisting of 41 complexes, our method has overall precision and recall values of 63% and 57%. On Benchmark v4.0, our method has overall precision and recall values of 55% and 56%. On CAPRI targets, our method has overall precision and recall values of 52% and 55%.

Research Area(s)

  • Hexagon structure, Neighborhood information, Protein–protein interface