Quantification of Entropy-Loss in Replica-Averaged Modeling.

Averaging across multiple replicas provides a straightforward and rigorous approach to employ averaged experimental data as restraints in molecular simulations. One significant practical obstacle is to optimally choose the number of replicas, N. Here, we describe a statistical method to estimate the intrinsic entropy-loss associated with modeling some data using N replicas, from an unbiased simulation. We discuss how having such a measure at hand may be used to assess N optimally.