Abstrait

Graph kernels and applications in protein classification

Jiang Qiangrong, Xiong Zhikang, Zhai Can

Protein classification is a well established research field concerned with the discovery ofmoleculeÂ’s properties through informational techniques. Graphbased kernels provide a nice framework combining machine learning techniques with graph theory. In this paper we introduce a novel graph kernel method for annotating functional residues in protein structures.Astructure is first modeled as a protein contact graph, where nodes correspond to residues and edges connect spatially neighboring residues. In experiments on classification of graphmodels of proteins, themethod based onWeisfeiler- Lehman shortest path kernel with complement graphs outperformed other state-of-art methods.