Multiscale modeling of pH-dependent biomolecule behavior
EMSL Project ID
60155
Abstract
Electrostatic interactions between charged biomolecules and the surrounding aqueous electrolyte solvent play a variety of important roles determining biomolecule structure and function. For studies in which fully atomistic molecular dynamics (MD) simulations with explicit solvent are impractically resource intensive, a simple continuum model based on macroscopic dielectric theory is often used. However, this model oversimplifies the behavior of the first shell of solvent molecules surrounding the biomolecule, leading to significant errors. We have recently corrected the model using a multiscale perturbation, yielding unprecedented accuracy. The improved model was originally developed to address errors in protein titration curves, and we have recently published an open-source GPU-accelerated simulation code that is scalable to large proteins. The NIH-funded APBS software (developed at PNNL) has extensive functionality to calculate protein pKas and titration curves, so we propose to test an integration where the APBS titration code calls our GPU-accelerated improved model, rather than calling the existing internal APBS simulation codes. We are requesting time on Tahoma so that the calculations can take advantage of the GPUs, and also for some all-atom MD simulations which may be needed (we have prior experience using NAMD for these purposes). After implementing the integration, we will validate it by setting parameters so that our simulation code reproduces the standard existing model, and compare our results to “traditional” APBS calculations of titration curves. We will then calculate titration curves for peptides and proteins using the SLIC model with parameters that have been optimized previously. We will also assess the impacts of salt concentration. The final stages of this effort will involve deploying the integration to the APBS web server (following appropriate reviews), incorporating the new model into the documentation, and public release of the new version of the APBS software. This work will advance capabilities to predict how the behavior of proteins and other biomolecules change as a function of the composition and temperature of their surrounding solvent environment, including pH and dissolved salt species. The continuum theory underlying the multiscale model is widely used in quantum-mechanical studies of molecules in solution, e.g. in NWChem and CP2K, and therefore the proposed work also supports future improvements to quantum-mechanical calculations.
Project Details
Start Date
2021-07-13
End Date
2023-09-30
Status
Closed
Released Data Link
Team
Principal Investigator