High-performance computing has a great potential to provide significant benefits for investigating biological systems in areas where experimental measurements cannot be made. These systems often present large modelling problems with many coupled subsystems, such as when studying communities of microbial cells. Understanding cell communities is of substantial interest due to their significant economic and societal impacts in industrial bioreactors and the complex community structures effecting plant growth in nature. However, investigating these microbial communities with realistic models can rapidly exceed traditional computational capabilities.
BMX (Biological Modelling and interface eXchange) is a software system developed for the high-performance modelling of large cell communities by utilizing hybrid parallelism with Graphics Processing Unit (GPU) acceleration. BMX builds upon the AMReX (Adaptive Mesh Refinement for the eXascale) adaptive mesh refinement package to efficiently model cell colony formation under realistic laboratory conditions. Using simple test scenarios with varying nutrient availability, BMX is capable of reproducing observed behavior of bacterial colonies on realistic time scales, demonstrating the efficient application of high-performance computing to cell community modelling.
High-fidelity simulations enabled by high-performance computing will allow for unprecedented predictive power of molecular level processes that are not amenable to experimental measurement. Many of these molecular level processes underly changes in soils and plant growth that are affected by climate change.
Understanding complex environments, such as microbial communities, is important because the molecular processes that tie together fungi and bacteria also provide nutrients in the soil for plant growth. With a rapidly changing environment due to climate change, understanding the microbial interactions that enhance plant growth is essential for environmental resiliency and sustainability. But interrogating these molecular processes at high resolution through experimentation is generally not feasible. Modeling and simulation can provide a window on these important processes, but realistic models that capture the physics of the processes are required. Modeling the physics accurately at the relevant scales in order to obtain actionable results is computationally demanding.
BMX was developed to produce physically realistic, large-scale simulations of 3D microbial community growth within real time frames. To this goal, BMX utilizes Central Processing Units coupled with Graphics Processing Unit acceleration using the AMReX package to provide a portable, high-performance simulation suite. Additional capabilities were added to the AMReX base to handle the dynamics of cells with the inclusion of associated interaction force laws and cellular metabolic models. Complex biological systems, such as biofilms and bioreactor systems, can include multiple cell species and chemical reactions and cell-cell signaling in the environment. BMX is a significant tool for understanding how microbial interactions impact the environment.
William R. Cannon, Pacific Northwest National Laboratory, email@example.com
This work was supported by the Department of Energy’s Biological and Environmental Research program through a contract to the Pacific Northwest National Laboratory.
Published: September 26, 2023
Palmer, B. J., A. S. Almgren, C. G. M. Johnson, A. T. Myers, W. R. Cannon. 2023. “BMX: Biological modelling and interface exchange.” Sci Rep 13, 12235. DOI: 10.1038/s41598-023-39150-1