“…Since the early implementations of this approach in the case of the replica-averaged NOE distance restraints, 13,15,16 a variety of restraining algorithms, including simultaneous time and ensemble averaging, 35 have been developed for an array of experimental observables measured for native, transition, intermediate, and unfolded states. 17,[36][37][38][39][40][41][42] One of the challenges in using replica-averaged structural restraints in molecular dynamics simulations, however, is that while the experimentally-determined average values are known, the underlying distribution (i.e., ensembles of structures) from which they come are not. 19,37 There is thus an ambiguity arising from the fact that infinitely many distributions may exist with the given set of average values.…”