Critical evaluation of the research docking program for the CASP2 challenge

The binding positions of six small-molecule ligands in their complexes with target proteins were predicted using our Research docking program for the CASP2 challenge. Research uses a Monte Carlo procedure with pairwise energies and allows for the conformational searching of ligand torsional space. We were able to predict 2 of the 5 noncovalent complexes within 2 A root-mean-square (RMS) of the experimental structures as ranked by interaction energy or by a score calculated using our interaction evaluation program, Out-rank. In addition, for 4 of the 5 noncovalent structures we found a docking within 2 A RMS of the experimental structure within the top 20 dockings as ranked by energy. The main limitation in our approach is in the ability of the energy function and Outrank to discriminate among the lowest energy dockings. On the other hand, our success in exploring the multi-dimensional docking space of position, orientation and conformation is encouraging.