The Designability of Protein Structures: A Lattice-Model Study using the Miyazawa-Jernigan Matrix

TL;DR

This study investigates the designability of lattice protein structures using the Miyazawa-Jernigan matrix, comparing results with hydrophobic-polar models, and finds consistent patterns of highly designable structures across models.

Contribution

It extends previous hydrophobic-polar models by using all 20 amino acids with empirical interaction potentials to analyze protein structure designability.

Findings

Highly designable structures are consistent across models.

Sequences for highly designable structures show increased thermodynamic stability.

Qualitative agreement between HP and MJ models in designability patterns.

Abstract

We study the designability of all compact 3x3x3 and 6x6 lattice-protein structures using the Miyazawa-Jernigan (MJ) matrix. The designability of a structure is the number of sequences that design the structure, i.e. sequences that have that structure as their unique lowest-energy state. Previous studies of hydrophobic-polar (HP) models showed a wide distribution of structure designabilities. Recently, questions were raised concerning the use of a 2-letter (HP) code in such studies. Here we calculate designabilities using all 20 amino acids, with empirically determined interaction potentials (MJ matrix), and compare with HP model results. We find good qualitative agreement between the two models. In particular, highly designable structures in the HP model are also highly designable in the MJ model--and vice versa--with the associated sequences having enhanced thermodynamic stability.