Fragmentation and Moving Domain Methods for the Computation of Electrostatic Potentials in Proteins

Date of Completion

January 2011


Chemistry, Biochemistry|Chemistry, Physical




This study presents several divide and conquer type algorithms to compute electrostatic potentials in large macromolecules such as proteins. One objective of this thesis concerns new developments of the Moving-Domain QM/MM (MOD-QM/MM) method: computational protocol that calculates quantum mechanically derived charges for a large molecular system. New developments in MOD-QM/MM include improved domain partitioning schemes and electrostatic potential charge fitting methods. The effect of scaling partial charges near the QM/MM boundary, the influence of the size of the QM domain and the basis set are also investigated. The effect of including polarized point charges in calculations is investigated in the context of metalloprotein-ligand docking and tryptophan absorption energy calculations. A second objective concerns developing a fragmentation-based method in the context of continuum electrostatic solvation via a conductor like screening model (COSMO). The new solvation model, termed Density Domain Fragmented COSMO (DDF-COSMO), provides a scalable approach to calculate solvation energies of large molecular systems. Due to its intrinsic fragmentation scheme, DDF-COSMO can be parallelized in a straightforward manner. The DDF-COSMO solvation model is integrated with the Moving-Domain QM/MM method via a double iterative scheme to include both self-polarization and the solvent effect. ^