Biomedical Engineering Reference
In-Depth Information
Fig. 6
All-atom protein model. (
a
) Schematic diagram for the all-atom protein model. Only
two consecutive residues are shown. The
solid thick lines
represent the covalent and the peptide
bonds. The
thin dashed lines
denote the effective bonds that are needed either to fix the bond
angles, model the side chain dihedral angles, or to maintain the planarity of the peptide bonds.
(
b
) Parameterization of the bonded interactions for representative atom pairs. The first column
shows the distribution of the distances in serine between N
O
”
, respectively.
The second column shows the corresponding histogram for the distribution of each atom pair. The
third column shows the resulting constraint potentials schematically. For bonds (e.g., N
C
'
; N
C
“
,andN
C
'
)and
bond angles (e.g., N
C
“
), the left and right boundaries of the constraint potential correspond to
d
C
, respectively. Here, d is the average length and is the standard deviation
of the distance distribution. (
c
) Parameterization of nonbonded interactions in all-atom DMD.
The continuous
red line
corresponds to the van der Waals and solvation interaction between two
carbon atoms. The
black step
function is the discretized potential for DMD. (
d
) A schematic for
the hydrogen bonding interaction between hydrogen H
i
and acceptor A
j
.AtomD
i
is the donor
and X
j
is the heavy atom directly bonded to A
j
. Besides the distance between the hydrogen and
the acceptor d
HA
, we also assess the auxiliary distances d
DA
(distance between atoms D
i
and A
j
)
and d
HX
(distance between atoms H
i
and X
j
)
and d
In order to accurately represent nonbonded interactions, we discretized the con-
tinuous Medusa force field [
34
], in which the VDW and solvation interactions are
included. VDW interactions use the standard Lennard-Jones potential, and solvation
interactions are modeled by the Lazaridis-Karplus (LK) solvation model [
41
],
which is expressed as the sum of pairwise distance-dependent effective solvation
energies (EEF1). The discrete potential functions mimic the continuous potential
E
ij
.d /
E
VDW
ij
.d /
C
E
LK
ij
.d / by capturing the attractions and repulsions while
using a minimal number of steps (Fig.
6
c). By trial and error in test simulations, we
adopted the following discretization protocol: (1) we choose an interaction range of
D
Search WWH ::
Custom Search