Our experiments show that the descriptor is effective in improving docking predictions when considered as one of the terms in the scoring function. The results on the ZDOCK benchmark MC1568 datasets have been compared with the other available shape-based docking algorithms, ZDOCK, Context Shapes, PatchDock, and HEX. in the ZDOCK benchmark 2.0 dataset. In 74% of the bound docking predictions, our method was able to find a near-native solution (interface C-RMSD 2.5 ) within the top 1000 ranks. For unbound docking, among the 60 complexes for which our algorithm returned at least one hit, 60% of the cases were ranked within the top 2000. Comparison with existing shape-based docking algorithms shows that our method has a better performance than the others in unbound docking MC1568 while remaining competitive for bound docking cases. == Conclusion == We show for the first time that the 3D Zernike descriptors are adept in capturing shape complementarity at the protein-protein interface and useful for protein docking prediction. Rigorous benchmark studies show that our docking approach has a superior performance compared to existing methods. == Background == Protein-protein interactions are a pivotal component of many biological processes and mediate a diverse variety of functions that include signal transduction, antibody-antigen complex, transport, and gene expression regulation, to name a few [1-4]. Although detailed structural information of a protein complex is necessary MC1568 for the understanding of the underlying mechanisms, such details are often difficult to obtain through traditional experimental means (X-ray crystallography, NMR). To this end, a number of computational protein docking methods have been developed [5,6], which employ various schemes for describing both geometric and energetic complementarity at the docking interface and explore the docking LPA antibody conformational space [7-11]. Although significant efforts have been made to develop protein docking prediction methods, recent Critical Assessment of PRediction of Interactions (CAPRI) [12] experiments have shown that plenty of room for improvement still remains. It is often discussed that one of the weaknesses of current docking methods is that they are not fully capable of handling the inherent flexibility of protein chains, which induces varying degrees of conformational change upon docking [13,14]. MC1568 Indeed, accounting for full flexibility of proteins is still a very challenging task for any computational method, especially for cases where the main-chain of a protein undergoes substantial conformational change (a root mean square deviation (RMSD) of a few Angstroms) upon docking. Such large fluctuations of loop regions or domain motions associated with MC1568 several proteins is difficult to predict even using current molecular dynamics simulation [15] orab initioprotein structure prediction approaches [16-18]. Therefore, some of the earliest works on protein docking treat interacting proteins as rigid-bodies using geometrical complementarity at the docking interface to guide the search for putative binding modes [9,19,20]. The importance of effective methods of capturing geometrical information has been recently emphasized because shape complementarity forms the basis of most docking schemes, including those that accommodate a reasonable amount of conformational change. This can be attributed to the following reasons: (1) There are many protein complexes which undergo only subtle conformational change upon docking, which are practically well approximated by rigid-body docking methods [21]. Enzyme-inhibitors, an important class of protein complexes, fall into this category. (2) Flexibility of the protein can be handled to some extent by introducing a degree of softness in the surface representation [22-24] or by applying energy optimization [15] or side chain conformation search. (3) To account for chain flexibility in the docking prediction, pair-wise docking of simulated conformational ensembles of the interacting proteins can be performed, while treating each pre-determined conformation as a rigid.