Figure 4.

Systematic benchmark studies of empirical force fields to improve their performance in folding simulations are now more frequently appearing in the literature.^{94, 169–174} (A) Various force fields were tested by Lindorff-Larsen et al.⁹⁵ and a score based on the performance of the force field against the chosen model systems was devised such that the lower score indicates better agreement with experimental data. The plot shows the improvement of force fields over time. (B) Piana et al.⁹⁶ used four different force fields to fold villin headpiece. Although all simulations arrived at the correct native state, the folding mechanism depended on the force field used. The panel shows that the flux through different reaction pathways (123 etc. is the order in which the three helices of villin headpiece form) is a function of the force field that was used.