Network Hamiltonian Models Provide Access to Scalable Protein Aggregation Simulations

Aggregation of γD-crystallin proteins in the eye lens are a known cause of both genetic and age-related cataract disease. Models of these aggregates are difficult to produce due to the size and complexity of traditional Molecular Dynamics (MD) simulations. In this paper, published in The Journal of Physical Chemistry B, NCASD member Liz Diessner and PI Carter Butts circumvent the complexities of all-atom MD simulations by applying a topological coarse-grain solution in the form of Network Hamiltonian models (NHMs). These models make use of Exponential-family Random Graph Models (ERGMs) to recapitulate the results of aggregation models performed using MD simulated structures, but only using information about the rates of contact between individual proteins. By reducing the amount of information needed for the model from all-atom positions and momenta to simple contact metrics between entire proteins, NHMs provide a computational approach for modeling aggregation systems that can scale the system size by orders of magnitude compared to those achieved by traditional MD. Such scalability allows researchers to probe the impact of system size on aggregation and examine underlying structure of large protein aggregates, in addition to providing the potential for exploration into phenomena such as liquid-liquid phase separation (LLPS) and phase transitions.