A new implicit solvent model for protein–ligand docking

Abstract

A new implicit solvent model for computing the electrostatics binding free energy in protein-ligand docking is proposed. The new method is based on an adaptation of the screening coulombic potentials proposed originally by Hassan et al. (J Phys Chem B 2000;104:6490-6498). In essence, it relies on two basic assumptions; (i) solvent screening can be accounted for by means of radially dependent sigmoidal dielectric functions and; (ii) the effective atom Born radii can be expressed only as a function of the exposed atom surface. Parameters of the model other than radii and charges are generic. These were optimized for a dataset of 826 protein-ligand complexes, comprising both X-ray complexes for 23 receptors as well as decoys generated by docking computations. We show that the new model provides satisfactory results when benchmarked against reference values based on the numerical solution of the Poisson equation, with a root mean square error of 4.2 kcal/mol over a range of approximately 40 kcal/mol in electrostatics binding free energies, a cross-validated r2 of 0.81, a slope of 0.97, and an intercept of 1.06 kcal/mol. We show that the model is appropriate for ligands of different sizes, polarities, overall charge, and chemical composition. Furthermore, not only the total value of the electrostatic contribution to the binding free energy, but also its components (coulombic term, receptor desolvation, and ligand desolvation) are reasonably well reproduced. Computation times of approximately 0.030 s per pose are obtained on a single processor desktop workstation.

Discrepancy in the near‐solute electric dipole moment calculated from the electric field

“…p E (r) was derived from X dip (r; h) in (17), not X pure dip (r; h) in eq. (18) or X pure E (r; h) in eq. (19).…”

“…Ignoring the orientation-dependent U ele;reac mean (r; h), X dip (r; h) was further approximated as X pure dip (r; h) (18). To understand the performance under these approximations, X pure dip (r; h), which was calculated from E(r) using eq.…”

“…X pure dip (r; h) and X pure E (r; h) calculated from the electric fields using eqs. (18) and (19), respectively, were compared with X(r; h), X mean (r; h) and X pot (r; h), c mol (r) calculated from the ratio of p and (e 0 E) was compared with c mol (r!1), and p E (r) calculated from the electric field using eq. (23) was compared with p(r).…”

“…Water solvent contributes the following effects: (1) it shields the electric interaction between charged particles; (2) it buries the hydrophobic group of solutes; and (3) it induces the magnetic field. For these reasons, the method of calculating P to study the behavior of molecular solutes is important for predicting/explaining the function of the proteins, [3,17] computer-aided drug design, [18][19][20] and nanotechnology. [21] P can be decomposed into the product of the solvent density q, and the electric dipole moment p, as P ¼ qp.…”

VSDMIP: virtual screening data management on an integrated platform

“…Six service interfaces currently exist: 3D structure generation, conformational analysis, atomic charge calculation, filtering, VS, and electrostatic calculations. Several plugins have been developed to interface with CORINA 3.0.5 [11] (3D generation), ALFA [12] (conformational analysis), MOPAC 7 [13] (atomic charges), DOCK 3.5 [14] and FRED 2.2 [15] (filter interface), CDOCK [16] and Autodock 3.0.5 [17] (VS interface), and for DelPhi 4 [18] and ISM [19] (electrostatic calculations).…”