Minimal number and selection strategy of FRET-based restraints needed to reach a target RMSD

Determine the minimal number of Cα–Cα distance restraints and identify which specific amino acid pairs must be restrained in all-atom molecular dynamics simulations to achieve a specified root-mean-square deviation in Cα positions relative to a target protein conformation; evaluate how random restraint selection compares to structured selection methods, including normal mode analysis, largest Cα separation, largest change in pairwise separation, and linear discriminant analysis.

Background

Because most FRET studies yield inter-residue distances for only a small number of labeled amino acid pairs, an essential practical question is how many restraints, and which ones, are needed to drive restrained MD simulations toward a target structure with a given accuracy (RMSD).

The authors highlight that prior to this work, it was unknown both how few restraints could suffice and how selection strategies (random versus informed methods) compare in effectiveness. Their paper implements and assesses several selection methods across multiple proteins, but the general problem of establishing minimal restraints and optimal selection strategies remains a central methodological question in FRET-assisted structural modeling.