Molecular simulations have two related goals: the search for a global
energy minimum (enthalpy) and the generation of an equilibrated ensemble
(free energy). Recently, several new Monte Carlo methods have been proposed
which focus on the generation of the equilibrium distribution of states,
rather than just a single, minimum enthalpy conformation. This view is
of particular importance when studying the properties of short, helical
peptides which do not exhibit simple two-state (random coil-> fully helical)
behavior, but rather are characterized by broad transitions, indicating
the presence of substantially populated intermediate states over a wide
range of temperatures and denaturant conditions. In this work, the applicability
of the method of Generalized Ensembles to 17 residue homo and heteropolymers
is explored. This method employs the Metropolis algorithm with a modified
(non-Boltzmann) weight function, designed to overcome energy barriers and
to thus enhance the sampling of conformational space. The resulting distribution
(given by the normalized weight function) is re-weighted to give the Boltzmann-distributed
ensemble. We have found that while the concept of non-Boltzmann sampling
is promising, the proposed weight function is not an optimal solution for
all systems. We therefore conclude that a heuristic approach to determine
the weight function, while requiring a substantia increase in computing
time, may be the most viable method. |
