Abstract
Molecular dynamics (MD) simulations are essential for investigating complex bi- ological systems. Timestep selection is crucial for accuracy and efficiency, yet the common practice of using a 4 fs timestep with hydrogen mass repartitioning (HMR) and SHAKE for alchemical free energy (AFE) calculations requires further scrutiny. This study investigates the impact of timestep on energy drift and AFE calculations across three molecular systems: cAMP, ethane, and the protein-ligand system Tyk2 (emj 42 to emj 55), exploring timesteps from 0.5 to 4 fs. NVE simulations revealed a strong correlation between increasing timestep and energy drift, with and without HMR. Simulations without SHAKE were unstable at larger timesteps. Thermody- namic integration (TI) simulations in the NPT ensemble showed consistent dU dλ values for 0.5, 1, and 2 fs timesteps, but deviations up to 3 kcal/mol were observed at 4 fs in some cases. Statistical t-tests (p < 0.05) confirmed significant differences between the 4 fs timestep and the 0.5 fs reference, especially in aqueous solution. These findings, based on analyses across various λ states (0-1) and softcore treatments, highlight the importance of careful timestep selection. We recommend limiting the timestep to a maximum of 2 fs for accurate and reliable AFE calculations using HMR and SHAKE.
Supplementary materials
Title
Impact of Timestep on Energy Drift and Accuracy of Alchemical Free Energy Calculations
Description
The total energy drift plots are shown for two systems
Actions