Protein-Protein Docking Using Multi-layered Spherical Basis Functions

The prediction of the structure of protein-protein complexes is an important but difficult problem mainly due to the huge computational cost required to search a configuration space. In this paper, we propose a fast global search docking algorithm that uses our novel multi-layered spherical basis functions (MLSBFs) to efficiently search the six-dimensional space of a rigid-body model. In our approach, the three-dimensional space is divided into layers and the scalar fields in each layer, used for calculating scores, are expanded by the MLSBFs to get the coefficients that express the fields. The MLSBFs enable the efficient usage of coefficients to express fields and thus an efficient global search. Our experiments showed that our program was able to generate predicted structures within a few minutes using a single 2.4 GHz Pentium 4 processor, and succeeded in listing at least one near native structure, whose interface RMSD is less than 2.5 angstrom, within the top 100 in 5 out of 7 cases.