INTRODUCTION

Protein-protein interactions are crucial in many biological processes like signal transduction,intracellular trafficking,and immune all proteinprotein interactions,a significant portion is formed by symmetric homo-oligomers(Andre et al.2008;Goodsell and Olson 2000; Poupon and Janin 2010).According to the Protein Data Bank(PDB)(Berman et al.2000),more than one third of the proteins have some types of example,many transmembrane proteins like ion channels are formed by symmetric homooligomer symmetry of homo-oligomeric proteins is thought to be associated with many potential benefits like greater stability,reduced aggregation,and robustness to errors in synthesis (Andre et al.2008;Goodsell and Olson 2000).The interface between symmetric homo-oligomers is often the targeting site for regulating the biological processes (Petsalaki and Russell 2008).Therefore,determining the complex structure of symmetric proteins is important (Lensink et al.2016,2018).Theoretically,one can use a general protein-protein docking approach to predict the complex structure of symmetric homo-oligomers by docking one monomer against the other(Comeau et al.2004;de Vries et al.2010,2015; Torchala et al.2013; Tovchigrechko and Vakser 2006).However,such a general docking strategy is not efficient for symmetric one hand,the general protein-protein docking approach treats two interacting partners as different proteins and therefore often don’t generate the complex structures with strict symmetry; On the other hand,general protein-protein docking normally don’t consider the symmetry restraints during the docking process,and therefore is not computationally ,specialized protein-protein docking algorithms are needed for predicting the complex structure of symmetric protein homo-oligomers.

One important symmetry in proteins is cyclic symmetry (Cn),for which the oligomeric structure can be constructed by n consecutive rotations of 360°/n around a single rotational axis of one subunit (Andre et al.2008).Despite the importance of symmetric protein homo-oligomers,only a few algorithms have been developed for symmetric protein et a fast docking algorithm for cyclically symmetric complexes through local feature matching,which is referred to as SymmDock(Schneidman-Duhovny et al.2005).SymmDock constructs the symmetric homooligomer complexes by restricting the search to symmetric cyclic Weng group developed an FFT-based algorithm for symmetric protein-protein docking by restricting the search space with cyclic symmetry (M-ZDOCK) (Pierce et al.2005).Based on the symmetric protein complexes in the PDB,several web servers that use template-based methods like ROBETTA (DiMaio et al.2011),SWISS-MODEL (Biasini et al.2014),and GalaxyGemini (Lee et al.2013)have also been proposed to predict the homo-oligomeric addition,Ritchie and Grudinin presented a fast docking algorithm,which is named SAM,for predicting the symmetrical models of protein complexes with arbitrary point group symmetry through a spherical polar FFT-based algorithm (Ritchie and Grudinin 2016).Very recently,the Seok group has developed a combination modeling approach,GalaxyHomomer,for homo-oligomer structure prediction from a monomer sequence or structure by template-based modeling if homologous complexes are available in the PDB or ab initio docking (Baek et al.2017).

However,despite the significant progress in the development of symmetric docking algorithms,there is still much room in improving the docking ,we have developed a new pairwise shapebased scoring approach to consider long-range interactions(LSC)of protein atoms by an exponential form in FFT-based protein-protein on general protein-protein complexes,our LSC approach showed a significant advantage over the traditional grid-based method (Yan and Huang 2018).Extending the LSC approach to symmetric complexes,we have here developed a fast ab initio docking approach for the symmetric docking of homo-oligomers with Cnsymmetry by an FFT-based search algorithm with LSC,which is referred to as CHDOCK.

RESULTS AND DISCUSSION

Comparison with other programs

We have tested our symmetric docking algorithm CHDOCK on the bound and unbound structures of our symmetric protein docking benchmark of 212 Cntargets (Yan and Huang 2019).Table 1 lists the success rates of CHDOCK in binding mode predictions for bound and unbound docking on the 212 cases with Cnsymmetry when the top 1,10,and 100 predictions are corresponding results are also shown in comparison,Table 1 and Fig.1 also give the corresponding results of three other Cnsymmetric docking algorithms,M-ZDOCK (Pierce et al.2005),SymmDock (Schneidman-Duhovny et al.2005),and SAM(Ritchie and Grudinin 2016),on this benchmark,in which the same clustering criteria have been applied to their final binding modes during the calculation of success rates.It can be seen from Table 1 and Fig.1 that CHDOCK obtained a significantly better performance than the other three docking methods for bound docking and achieved a success rate of 55.19%,72.17%,and 90.57% for top 1,10,and 100 predictions,respectively,in comparison to those of 45.76%,65.09%,and 89.15%for M-ZDOCK,38.85%,54.25%,and 84.91% for SAM,and 16.04% 31.60%,and 67.45% for SymmDock.