Statistical mechanics of biomolecular condensates via cavity methods

TL;DR

This paper develops a cavity method-based statistical mechanics approach to model biomolecular phase separation, successfully applying it to complex systems and validating with simulations and experiments.

Contribution

It introduces a cavity method framework for accurately modeling biomolecular condensates beyond mean field approximations.

Findings

Agreement with lattice model simulations

Successful application to ternary systems

Validation with experimental coacervate data

Abstract

Physical mechanisms of phase separation in living systems can play key physiological roles and have recently been the focus of intensive studies. The strongly heterogeneous and disordered nature of such phenomena in the biological domain poses difficult modeling challenges that require going beyond mean field approaches based on postulating a free energy landscape. The alternative pathway we take in this work is to tackle the full statistical mechanics problem of calculating the partition function in these systems, starting from microscopic interactions, by means of cavity methods. We illustrate the procedure first on the simple binary case, and we then apply it successfully to ternary systems, in which the naive mean field approximations are proved inadequate. We then demonstrate the agreement with lattice model simulations, to finally contrast our theory also with experiments of coacervate formation by associative de-mixing of nucleotides and poly-lysine in aqueous solution. In this way, different types of evidence are provided to support cavity methods as ideal tools for quantitative modeling of biomolecular condensation, giving an optimal balance between the accurate consideration of spatial aspects of the microscopic dynamics and the fast computational results rooted in their analytical tractability.