GROMACS in the Cloud: A Global Supercomputer to Speed Up Alchemical Drug Design

Abstract

We assess costs and efficiency of state-of-the-art high-performance cloud computing and compare the results to traditional on-premises compute clusters. Our use case is atomistic simulations carried out with the GROMACS molecular dynamics (MD) toolkit with a particular focus on alchemical protein–ligand binding free energy calculations. We set up a compute cluster in the Amazon Web Services (AWS) cloud that incorporates various different instances with Intel, AMD, and ARM CPUs, some with GPU acceleration. Using representative biomolecular simulation systems, we benchmark how GROMACS performs on individual instances and across multiple instances. Thereby we assess which instances deliver the highest performance and which are the most cost-efficient ones for our use case. We find that, in terms of total costs, including hardware, personnel, room, energy, and cooling, producing MD trajectories in the cloud can be about as cost-efficient as an on-premises cluster given that optimal cloud instances are chosen. Further, we find that high-throughput ligand-screening can be accelerated dramatically by using global cloud resources. For a ligand screening study consisting of 19 872 independent simulations or ∼200 μs of combined simulation trajectory, we made use of diverse hardware available in the cloud at the time of the study. The computations scaled-up to reach peak performance using more than 4 000 instances, 140 000 cores, and 3 000 GPUs simultaneously. Our simulation ensemble finished in about 2 days in the cloud, while weeks would be required to complete the task on a typical on-premises cluster consisting of several hundred nodes.

Enhanced sampling for lipid-protein interactions during membrane dynamics

“…The predictive capabilities of computational simulations have reached an interesting stage in the biomedical sciences. Improved simulation packages (Brooks et al, 2021;Suh et al, 2022), better force-fields (Klein et al, 2021;Souza et al, 2021;Cruz-León et al, 2021;Yungerman et al, 2022), unprecedented supercomputer power (Yamazaki et al, 2021;Kutzner et al, 2022) and creative sampling techniques (Gilabert et al, 2019;Bonati et al, 2021) have boosted the study of exceptionally complex biological problems (Mosalaganti et al, 2022;Lotz and Dickson, 2018). More studies combining theoretical approaches, computer simulations, and experiments are currently envisioning new possibilities (Sica and Smulski, 2021;Bernetti and Bussi, 2021;Miguel et al, 2021;Quevedo et al, 2019;Saen-oon et al, 2015).…”

Predicting antibiotic resistance in complex protein targets using alchemical free energy methods

“…Significant speed up could however be achieved by use of updated GROMACS software, where yearly updates have been shown to increase ns/day performance on a range of computational resource. 39,40 It is not in doubt that how the structure and dynamics of a protein change upon mutation contains valuable information that can, in theory, be used to predict whether individual mutations confer antibiotic resistance. Although RBFE may not (yet) be an appropriate tool for resistance prediction in the two complex systems we studied, several alternative routes exist.…”

“…This makes simulations of these size prohibitively computationally intensive using the software and compute that we employed. Significant speed up could however be achieved by use of updated GROMACS software, where yearly updates have been shown to increase ns/day performance on a range of computational resource 39,40 …”

Chemical Space Exploration with Active Learning and Alchemical Free Energies

“…Alchemical free energy calculations − based on first principle statistical mechanics may serve as an optimal input for such AL applications. While computationally demanding, nowadays these calculations are readily accessible even at large scale: the predictions for hundreds to thousands of ligands can be obtained in a matter of days. , Also, the accuracy of alchemical predictions draws close to the experimental measurements. − Therefore, using these calculations as an oracle to construct ML models could allow describing binding affinities of large chemical libraries with high accuracy, while only a small fraction of the library needs to be evaluated with the computationally expensive alchemical method.…”