Enabling Detailed, Biophysics-Based Skeletal Muscle Models on HPC Systems.

Area of Science:

• Computational biology
• Biophysics
• Scientific computing

Background:

• Realistic multi-scale biophysics models require significant computational resources.
• Existing biomedical simulation environments prioritize flexibility, limiting scalability for large-scale computations.
• Skeletal muscle modeling presents unique challenges due to its complex chemo-electromechanical properties.

Purpose of the Study:

• To upgrade an existing skeletal muscle simulation framework to achieve massive parallel scalability.
• To investigate and implement improvements in modeling, algorithms, and implementation for enhanced performance.
• To develop a novel visualization environment for handling large-scale simulation data.

Main Methods:

• Utilized a detailed biophysics-based, chemo-electromechanical skeletal muscle model.
• Employed the OpenCMISS open-source software library for framework development.
• Investigated various modeling, algorithmic, and implementational strategies for parallelization.
• Integrated the MegaMol framework for advanced data visualization.

Main Results:

• Achieved significant improvements in both numerical and parallel scalability.
• Enhanced the simulation framework to run on massively parallel architectures.
• Improved overall runtime by a factor of up to 2.6.
• Demonstrated good scalability on up to 768 compute cores.

Conclusions:

• The developed approach successfully transforms a moderately parallel simulation framework into a massively parallel one.
• The enhanced framework enables larger problem sizes and efficient utilization of numerous parallel processes.
• The novel visualization environment effectively handles large datasets generated by high-resolution simulations.
• This work facilitates more comprehensive and computationally intensive studies of skeletal muscle biophysics.