Improvement Of Protein-Protein Docking And Application With Experiment Data

The development of computational docking methods to predict the structures of protein-protein complexes is important. One widely used technique of docking is the fast Fourier transform (FFT) method which can globally search the rotational and translational space of ligand quickly. Based on the FTDOCK program, we changed the parameters according to the statistical analysis of a decoy set, and altered the strategy of keeping highest score conformations in each rotational orientation to increase the success rate and efficiency of sampling. According to the results, our method is an improvement relative to the original FTDOCK program on aspects of average hit number and success rate. Furthermore, we filtered the retained complex structures with energy of Ambiguous Interaction Restraints (AIRs) which are used in HADDOCK. The result showed that the additional‘experimental’information was very effective to find near-native conformations. All the work was tested on the protein-protein docking benchmark 2.0. And three complexes with experiment data from paper of HADDOCK were also docked to test our method.In Chapter 3, hCLP and F-actin were docked using docking program HEX4.2. The retained conformations were filtered according to the experiment data, and clustered subsequently. The most likely model of hCLP-F-actin was finally given.