Monte Carlo simulations reveal limitations of the single-reaction-coordinate picture

The folding dynamics of biopolymers can be studied using a constrained random walk

of the configurational degrees of freedom over an appropriately chosen energy landscape.

The simplest toy statistical mechanical models involve a limited set of interacting

molecular elements living on the sites of a lattice, arranged so that they trace out

a self-avoiding walk. We consider a slightly more realistic situation in which elements

of the chain backbone move in the continuum and are subject to nearest-neighbor bond-bending costs

and to long-range interactions that account for excluded volume effects, hydrophobic attraction,

and electrostatics. We have developed local and global update schemes that,

in combination with parallel tempering, produce an ergodic and efficient exploration

of the phase space. These computational tools are put to work on real protein sequences

in order to highlight the limitations of the conventional analysis of pulling experiments that

assumes projection onto a single reaction coordinate.