Docking methods conformation space

Typically, a GA run will take several minutes, and most of the publications in this field suggest trying several runs to impart a reasonable confidence that conformational space has been adequately sampled. In addition, several researchers have included a desolvation term into their calculations, which further decreases their speed. q flexible GA-based docking procedure can take 100-1000 times longer than a traditional rigid DOCK run. This time constraint limits the utility of the GA-based flexible docking methods for the analysis of large databases. 

While the steric method described above is very efficient, in many cases, geometric criteria alone are insufficient to correctly dock the two molecules. This is especially true when the stmcmre of the receptor is of poor quality or a ligand molecule is relatively small so that shape complementarity is insufficient to specify the correct conformation. To overcome this problem, we decided to build a statistical potential that could be used for additional evaluation of the quality of the match. In order to build the potential, we defined 20 general atom types and built the contact statistics on the basis of the structures of known complexes available in the PDB [171]. After projection of the two molecules onto the grid, every cube is additionally labeled with the properties defined by the atom types that were projected onto it. Once the approximate representation of the system is ready, the best match of these two cube-clusters is determined by exhaustive scanning over the six-dimensional conformational space of the three relative translations and the three rotations. Calculating the value of the correlation function between these two sets of cubes and the value of the potential function, the quality of the particular ligand-receptor orientation is scored. 

---