A flexible ligand docking protocol based on evolutionary algorithms is inve
stigated. The proposed approach incorporates family competition and adaptiv
e rules to integrate decreasing-based mutations and self-adaptive mutations
to act as global and local search strategies, respectively. The method is
applied to a dihydrofolate reductase enzyme with the anticancer drug methot
rexate and two analogues of antibacterial drug trimethoprim. Conformations
and orientations dosed to the crystallographically determined structures ar
e obtained, as well as alternative structures with low energy. Numerical re
sults indicate that the new approach is very robust. The docked lowest-ener
gy structures have root-mean-square derivations ranging from 0.67 to 1.96 A
ngstrom with respect to the corresponding crystal structures. (C) 2000 John
Wiley & Sons, Inc.