ABSTRACT
Computer simulations of biological molecules have come a long
way since the initial 5 ps molecular dynamics (MD) simulations of
bovine pancreatic trypsin inhibitor in vacuum using an empirical
potential energy function (PEF) [1] almost 40 years ago. Recently,
one of the huge stumbling blocks in MD simulations of biological
problems, namely, sampling, has been greatly alleviated through
special purpose computers for MD simulations, which currently can
simulate small proteins in water on a millisecond timescale [2].
However, these simulations bring focus on another problem: the
accuracy of the potential energy functions. For instance, simulations
of multiple folding and unfolding events of the villin headpiece in
aqueous solution using two different parameter sets for the protein
and two different parameter sets for the water give the same
final folded state but different folding pathways [3]. In addition,
simulations of several proteins using different parameter sets
indicate that the radius of gyration of the unfolded state is about
half of the experimental value and close to that of the simulated
folded state [4]. Moreover, these simulations indicate that not only is
the accuracy of the biomolecular PEF important, but also the water
PEF. In other words, although the biomolecule may be the subject
of interest, the aqueous environment is necessary for the structure
and function of these molecules and so proper modeling of water is
essential.