Large FMM-Gromacs calculations for electrosprayed proteins
Abstract
A fast multipole method (FMM) for calculation of electrostatic interactions was recently implemented in the molecular dynamics software GROMACS, as an alternative to the commonly used particle mesh Ewald (PME) method. The attractiveness of FMM typically lies in its better scaling properties for highly parallel simulations, but in addition to this, it could be particularly beneficial for simulations of biomolecules in the gas phase, for two reasons: 1) FMM permits that systems are treated with open boundaries, instead of with periodic boundary conditions, which by construction is imperative with particle mesh Ewald method. This is a more realistic and accurate approach for gas-phase simulations. 2) It has been demonstrated that systems that are large and/or have inhomogeneous particle distribution benefit from significant performance enhancements with FMM. Gas-phase proteins fir that description, and in addition their electrostatic interactions are much more long-range because of the absence of a high-dielectric medium such as water, and FMM offers an efficient way to calculate them collectively.
We will use GROMACS with FMM to simulate proteins under electrospray conditions. These systems are gaseous and highly charged, offering great opportunity for performance enhancements in addition to accurate treatment of boundaries. We hope to achieve particular speed-ups for large and highly charged protein complexes, simulations of which are currently severely restricted by low computational efficiency. Simulations of electrosprayed proteins are routinely performed to guide interpretation of experimental data from mass spectrometry measurements. Our research group also uses simulations of this kind to inform in the development of new mass spectrometric techniques.
This proposal complements the NAISS Small project NAISS 2023/22-147, where we have managed to find a range of good simulation conditions. Here, we will first expand our investigation using a range of larger protein complexes to charter how the optimal simulation parameters vary for larger systems. Then we will apply the method to large protein complexes for which we have ongoing collaborations with experimentalist groups. First, we will simulate virus capsid assembly intermediates, which we have an ongoing project about and will have complementary data from experiments. We will also simulate a citrate synthase that forms a number of large multimers with a completely novel complex topology. For both these systems there are gas-phase data available or within reach with future experiments. Our efforts will potentially lead to considerably better gas-phase simulations of large macromolecules, as well as structural interpretation of ongoing experiments.
