Folding in two-dimenensional off-lattice models of proteins

TL;DR

This paper investigates two-dimensional off-lattice protein models, demonstrating that native states can be reliably identified and that Go-like native contact models facilitate effective folding, while other contacts hinder it.

Contribution

It introduces off-lattice 2D protein models with native states near on-lattice targets and evaluates folding properties of different interaction schemes.

Findings

Native states are reliably determined via Monte Carlo methods.

Rank-ordering contact energies often fails in off-lattice models.

Go-like models with native contacts show good folding behavior.

Abstract

Model off-lattice sequences in two dimensions are constructed so that their native states are close to an on-lattice target. The Hamiltonian involves the Lennard-Jones and harmonic interactions. The native states of these sequences are determined with a high degree of certainty through Monte Carlo processes. The sequences are characterized thermodynamically and kinetically. It is shown that the rank-ordering-based scheme of the assignment of contact energies typically fails in off-lattice models even though it generates high stability of on-lattice sequences. Similar to the on-lattice case, Go-like modeling, in which the interaction potentials are restricted to the native contacts in a target shape, gives rise to good folding properties. Involving other contacts deteriorates these properties.