Chemistry Reference
In-Depth Information
Fig. 8 A comparison of the NMR docking energy from AutoDockFilter to the rmsds between
the best docked ligand conformers and the experimental protein-ligand co-structure. An
improved correlation is observed for the docking of ligands to the bound form of the protein
(
circles
) compared to the apo-protein structure (
squares
). The
red
data points correspond to
AutoDockFilter docking results using experimental CSPs for staphylococcal nuclease (PDB-ID:
1EY0, 1SNC) [
180
-
182
]. The
yellow
data points correspond to a docking to the apo-structure of
acetylcholinesterase (PDB-ID:1ACJ, 1QIF) that resulted in a high rmsd. However, the inclusion
of side chain flexibility for residues in the ligand binding site resulted in an improved docking
and lower rmsd. (Reprinted with permission from [
121
], copyright 2008 by the American
Chemical Society)
Gonzalez-Ruiz and Gohlke describe a conceptual hybrid (QCSP-Steered
Docking) of the AutoDockFilter and the HADDOCK/LIGDOCK procedures, effec-
tively combining the best features of both methods [
160
]. AutoDock 3.0.5 was
modified to incorporate a new hybrid scoring scheme utilizing the DrugScore target
function [
164
] with an amended CSP energy function. Basically, AutoDock is used
to generate poses similar to AutoDockFilter, but when an energetically acceptable
pose is obtained, CSPs are calculated for the pose. The calculated CSPs are based
only on ring current effects [
165
] from aromatic rings in the ligand. A comparison
between the calculated and experimental CSPs is used to calculate an energy
violation. Instead of an absolute difference, a Kendall's rank correlation coefficient
is used to account for magnitude differences between the experimental and calcu-
lated CSP values. The pose with the lowest DrugScore and CSP energy is chosen.
Thus, QCSP-Steered Docking is as fast as AutoDockFilter, but allows for direct
refinement against the experimental CSPs like HADDOCK/LIGDOCK.
