binding regions) are required in order not to neglect the solu-
tions located on the borders of the chosen segments.- Normally, a cutoff used was 8 A ̊, whereas such a distance
between the receptor and the ligand was defined as 20–30 A ̊.
Such difference is needed to provide the ligand enough time to
sample its conformational space when targeting to the receptor
without perceiving any influence of the receptor on this
sampling. - Although normally there is enough time for a ligand to be
relaxed during the targeted MD step of a DMD run, for
consistency we recommend to use the same force field for the
minimization of the ligand as the one used in the DMD run. In
the original paper [14], we propose several initial orientations
of the ligand for different DMD runs. However, for each ligand
type and size, dependence of the results on the initial orienta-
tion can be different and, therefore, should be investigated.
According to our experience, the docking results obtained by
DMD using the protocol presented here do not depend on
ligand’s initial orientation for ligand of the size up to octameric
oligosaccharide. However, if this dependence is observed,
either several initial random orientations of ligand or a longer
targeted MD step should be used to avoid the bias originated
from a particular initial orientation of the ligand. - TheCatom should be chosen based on the following criteria:
(1) this atom should not essentially fluctuate in the MD simu-
lation (this information could either be obtained from the
previously run MD simulations of the receptor or could be
proposed based on the chemical type of the atom and its
belonging to secondary structure elements. In particular, the
backbone atoms fromα-helices andβ-sheets in a protein core
can be used since they are usually not too mobile); (2) the
vector drawn betweenLandCatoms should be roughly per-
pendicular to the surface comprising the predefined binding
region; (3) the surface of the sphere aroundCatom with the
radius equal to the distance fromCto the point of the surface
of the protein defined by this vector should have significant
overlap with the surface in the predefined binding region. The
last criterion is usually fulfilled for most global proteins unless
dramatic peculiarities of their surface geometry are observed. - Although 4 A ̊seem to be quite a short distance and a potential
source of artifacts, the ligand is targeted toward the receptor,
going “into” the octahedral solvated box. Therefore, this initial
distance to the solvent box boundaries does not affect the
results. Moreover, because of the big size of the box due to
the long initial distance between the receptor and the ligand,
an increase of the box size would lead to undesirable
452 Sergey A. Samsonov