Biomolecular Recognition: The Current Challenge - Dynamic Force Spectroscopy and Biomolecular Recognition - page 37

chemical considerations and experimental data. Using femtoseconds time steps, this
method currently allows the observation of objects as complex as proteins during
less than a microsecond, which is still insufficient to account for molecular inter-
actions. Calculations are performed with arrays of computers, using a few standard
software packages such as CHARMM [23], Gromos [40], Rosetta [48], or the freely
available Xplor-NIH package [176]. In the following sections, we shall first discuss
rapidly some limitations of the method and tricks suggested to progress. Then, we
shall describe a few selected examples to illustrate the kind of information that can
be obtained. More information can be found in aforementioned papers and standard
treatises [164].
1.5.3.1 Limitations and Technical Advances
As shown above, there is a need to extend the range of accessible domains to take
real advantage of computer simulations. Here are some noticeable points.
Long distance cut-off of interactions. Since most interatomic forces exhibit a rapid
decay with distance, it seemed reasonable to neglect interactions at distance higher
than some arbitrary cut-off. A useful point is that artefacts may be avoided by using,
for example, a smoothened cut-off [23].
Poisson-Boltzmann statistics. Due to the complexity of solvent effect, it was tempting
to account for water dielectric constant by considering individual molecules. Unfor-
tunately, this is highly time-consuming. It was suggested to use numerical solution
of Poisson-Boltzmann equation to calculate the potential on the surface of proteins
[88].
Protein flexibility. A major difficulty is the formidable number of configurations
available for proteins. Initial docking software thus treated proteins as rigid objects
[68] [179], which greatly increased calculation speed, but it is also an important lim-
itation, as discussed above. This was an incentive to elaborate clever research algo-
rithm to sample a sufficient variety of conformations within a reasonable amount
of time. Examples are the use of the so-called “genetic search algorithms,” which
allowed successful simulation of association of couples such as methotrexate and
dihydrofolate reductase, or galactose and L-arabinose-binding protein [100].
Coarse-grained potential. A major limitation of molecular dynamics is that the pas-
sage across significant energy barriers may require an excessive amount of time. This
difficulty may be overcome by smoothing barriers to allow a more rapid exploration
of more extensive configurations. In a later step, a more refined potential may be
used [133].
Brownian dynamics and hybrid simulation. Another way to overcome the time
limitation of molecular simulations consists of combining deterministic equations