Lagrangian Particle Methods for Biological Systems - Porous Media: Applications in Biological Systems and Biotechnology - page 255

of the surfactant was represented by one DPD particle, the hydrophobic tail
was represented by five particles, and the particles were connected by har-
monic springs. When the surfactant was dissolved in a solvent consisting of
water-like hydrophilic DPD particles, the surfactant spontaneously assembled
into a bilayer. The collective behavior of molecules forming the membrane is
properly retained while other unnecessary atomistic level details (not of inter-
est in the investigation [Venturoli and Smit 1999]) were discarded to minimize
the computational cost.
Later, DPD was applied to the dynamic behavior of biological membranes
(membrane damage, rupture, and morphology changes) that are exposed to
nonionic surfactants (Groot and Rabone 2001). In principle, biological pro-
cesses take place over a wide range of timescales, and simulations must be run
for long enough to allow phenomena such as phase transition, phase separa-
tion, and the uptake of surfactant by the biological system to occur. Therefore,
DPD provides a very promising computationally ecient mesoscale approach
to the simulation of biological systems.
The successful application of DPD to specific molecular systems depends
on the parameters associated with the interactions among the various types
of DPD particles that represent the distinct molecular species that consti-
tute the system. These interaction parameters may be obtained by matching
experimental and observed properties (Groot and Rabone 2001). For small
molecules that can be represented by a single DPD particle, or a group of
identical particles, the interactions between particles representing the same
chemical component may be obtained by matching the experimental and sim-
ulated compressibility, and solubilities are often used to calibrate the inter-
actions between DPD particles that represent different chemical components.
Many larger molecules are represented by a number of different particles,
and calibration of the particle-particle interactions is more challenging. Sim-
ple surfactants may often be represented by a “head” consisting of one or
more particles of type A and a “tail” consisting of one or more particles of
type B. In this case, neutron scattering and/or x-ray scattering may be used
to obtain information about the density profile of the head or tail groups
at a liquid/air interface, and this information may be used to (partially) cali-
brate the particle-particle interactions. MD simulations may be used to obtain
more detailed information, and this information may be used to calibrate or
validate DPD models (Groot and Rabone 2001). In more complex systems, a
larger number of distinct DPD particles is needed, and the number of particle-
particle interactions is n ( n +1) / 2, where n is the number of distinct types of
DPD particles, and this is a lower limit on the number of interaction param-
eters. Consequently, the challenge of calibrating DPD models grows rapidly
with the complexity of the system, and there is rarely, if ever, sucient rele-
vant experimental information. An alternative approach, largely under devel-
opment, is to base the DPD model on a MD model by means of a systematic
coarse-graining procedure. If DPD simulations are used to investigate generic
behaviors, there is no need for system-specific interaction parameters, and a