Improving the accuracy and efficiency of docking methods

Abstract

Computational methods for predicting macromolecular complexes are useful tools for studying biological systems. They are used in areas such as drug design and for studying protein-protein interactions. While considerable progress has been made in this field over the decades, enhancing the speed and accuracy of these computational methods remains an important challenge. This work describes two different enhancements to the accuracy of ClusPro, a method for performing protein-protein docking, as well as an enhancement to the efficiency of global rigid body docking. SAXS is a high throughput technique collected for molecules in solution, and the data provides information about the shape and size of molecules. ClusPro was enhanced with the ability to SAXS data collected for protein complexes to guide docking by selecting conformations by how well they match the experimental data, which improved docking accuracy when such data is available. Various other experimental techniques, such as NMR, FRET, or chemical cross linking can provide information about protein-protein interfaces, and such information can be used to generate distance-based restraints between pairs of residues across the interface. A second enhancement to ClusPro enables the use of such distance restraints to improve docking accuracy. Finally, an enhancement to the efficiency of FFT based global docking programs was developed. This enhancement allows for the efficient search of multiple sidechain conformations, and this improved program was applied to the flexible computational solvent mapping program FTFlex.