The Many Faces of Structure-Based Potentials: From Protein Folding Landscapes to Structural Characterization of Complex Biomolecules - Computational Modeling of Biological Systems - page 34

depend even less on the precise atomic energetic details and more on emergent
properties controlled by the geometry of their constituents. While all these systems
will have residual levels of frustration, the use of SBMs as a baseline is crucial to
partition the global properties, those largely dependent on structure, from the details
dependent on specific energetics.

2.2 Structure-Based Model as a Baseline

Simplified models have a long history of elucidating the organizing principles
governing complex systems. A key question is how sensitive a model is to its
underlying parameters. Determining the correct value for a parameter is often
equally important as understanding the sensitivity to perturbations in that parameter.
Since molecular geometry has a central influence on the motions leading to
molecular function, simplified models based on low free energy structures are a
natural starting point. The simplest models look at the normal modes of an energy
landscape created by replacing all short range interactions in a native structure by
Hookean springs [ 61 ]. These models can capture relevant rigid body motions. SBMs
provide an important generalization by allowing the possibility for “cracking,”
[ 24 , 25 , 40 , 68 ] allowing interactions to break and reform, since the springs are
replaced by short range potentials. Thus, SBM can capture motion on all scales
from native basin dynamics to unfolding.
The straightforward formulation of a structure-based potential allows for sen-
sitivity analysis of the force field parameters [ 69 ] and their simplicity makes
them extremely fast to compute. The force field is readily extensible allowing the
introduction of complicated effects to be explored parametrically. For example,
the effects of electrostatics can be explored by perturbative addition of Coulomb
interactions [ 4 , 14 , 35 ], or the effects of solvent probed by the perturbative addition of
desolvation barriers [ 12 ]. A crucial question in the protein folding field has been how
proteins manage to achieve such smooth energy landscapes, or equivalently, why do
AA empirical force fields and structure prediction schemes have difficulty achieving
the level of specificity seen in proteins? Using structure-based potentials with
AA geometries, we can begin to address this question. These models completely
partition energetic effects from geometric effects, and through careful investigation,
may discern to what extent energetics contribute to the apparent native specificity
in protein structure, folding, and function. While processes like the formation
of nonnative intermediates during folding [ 18 , 53 , 60 ] and protein misfolding are
clearly cases that perfectly funneled SBM will be unable to fully describe, through
adding complexity in a piecemeal fashion to a robust baseline model, a more
complete understanding of the interplay between geometry and energy in even these
complicated systems will result.