Metadynamics bias exchange

One of the problems in metadynamics is that the maximum number of CVs that can be used in practice is limited. In particular, it is very difficult to converge metadynamics simulations performed using more than three CVs. This limits the applicability of the method to cases where a few appropriate CVs are known a priori. However, if one is able to define a super-set of CVs, which is very likely to include the appropriate CVs, multiple metadynamics simulations can be conveniently combined by means of bias exchange metadynamics (BEMD).  

Using BEMD it is thus possible to replace an (unfeasible) metadynamics that would use ten or twenty CVs with ten or twenty weakly coupled one-dimensional metadynamics simulations and retain an accuracy that is determined by the replica containing the optimal CV. Moreover, the statistics of each of the replicas benehts from the exchange. Accordingly, the exploration of phase space by each replica is much higher compared to that of a simple MetaD. We hnally point out that in BEMD the statistics produced by all the replicas should be used to calculate the final free-energy landscape via a reweighting procedure. 

---

In this way the authors successfully produced a varied ensemble of secondary structures. More recently, these variables were used in conjunction with an extensive bias exchange MetaD (see Section Bias Exchange Metadynamics ) on VAL60 to sample, with atomistic resolution, plausible folds containing secondary structures well beyond the variety contained in the PDB database. 

F. Baftizadeh, P. Cossio, F. Pietrucci, and A. Laio, Curr. Phys. Chem., 2,79-91 (2012). Protein Folding and Ligand-Enzyme Binding from Bias-Exchange Metadynamics Simulations. 

Thanks to their complementarity, metadynamics and replica exchange can nonetheless be optimally combined.To do so one has to perform simultaneous metadynamics simulations and adjust the Metropolis criterion to include the bias potential... 

---