Accelerating atomic-level protein simulations by flat-histogram techniques

Abstract

Flat-histogram techniques provide a powerful approach to the simulation of first-order-like phase transitions and are potentially very useful for protein studies. Here, we test this approach by implicit solvent all-atom Monte Carlo (MC) simulations of peptide aggregation, for a 7-residue fragment (GIIFNEQ) of the Cu/Zn superoxide dismutase 1 protein (SOD1). In simulations with 8 chains, we observe two distinct aggregated/non-aggregated phases. At the midpoint temperature, these phases coexist, separated by a free-energy barrier of height 2.7 k B T . We show that this system can be successfully studied by carefully implemented flat-histogram techniques. The frequency of barrier crossing, which is low in conventional canonical simulations, can be increased by turning to a two-step procedure based on the Wang-Landau and multicanonical algorithms.

Distinct phases of free α‐synuclein—A Monte Carlo study

“…This bimodality provides a convenient partitioning of the simulated ensemble into high‐ and low‐energy subensembles, which we will call D and B, respectively. To facilitate crossing the large free‐energy barrier implied in Figure 1, we follow a two‐step simulation scheme based on flat‐histogram techniques 45…”

“…A detailed description of our simulation procedure can be found in Ref. 45. For a review of generalized‐ensemble techniques, see Ref.…”

Thermodynamics of amyloid formation and the role of intersheet interactions

“…For our thermodynamic simulations, we therefore use a Swendsen-Wangtype cluster move 40 and a flat-histogram procedure. [41][42][43]47 The cluster move is based on a stochastic cluster construction scheme. 40 The procedure …”

“…Our simulation procedure consists of three steps. 47 First, we estimate the density of states, g(E), by the Wang-Landau method 43 (an early variant of which was proposed in Ref. 48).…”

Dynamical traps in Wang-Landau sampling of continuous systems: Mechanism and solution

“…To date, WLS has been applied to a wide variety of systems ranging from random bond and random field systems [2][3][4], atomic fluid [5] and clusters [6], liquid crystals [7], polymers [6,8,9] and proteins [10][11][12], to lipid membranes in explicit solvent [13,14]. In MD and MC, one frequently encounters the problem that the system gets trapped in some local minimum due to the presence of large barriers on the energy landscape [11,15,16,29].…”