High Performance Adaptive Physics Refinement to Enable Large-Scale Tracking of Cancer Cell Trajectory.

Area of Science:

• Computational biology
• Biophysics
• Medical simulation

Background:

• Simulating cancer cell transport in the circulatory system requires high-resolution models of large fluid volumes, exceeding current supercomputing capabilities.
• Understanding metastatic spread necessitates detailed tracking of cancer cells within complex physiological environments.

Purpose of the Study:

• To develop a novel computational method for simulating cancer cell transport at cellular-scale resolution across large domains.
• To overcome the limitations of existing supercomputers in modeling fluid dynamics and cellular interactions for metastatic spread research.

Main Methods:

• Introduction of an adaptive physics refinement (APR) method integrating multi-physics and multi-resolution models.
• Leveraging a hybrid CPU-GPU approach for enhanced computational performance.
• Coupling finely resolved cellular-scale windows with coarsely resolved bulk fluid domains.

Main Results:

• Successful validation of the APR framework against fully resolved fluid-structure interaction methods.
• Implementation of performance optimization techniques, including latency hiding and memory bandwidth maximization.
• Demonstration of a robust and scalable framework for system-level simulations of cancer cell transport.

Conclusions:

• The developed APR method provides a powerful and efficient tool for simulating cancer cell dynamics.
• This computational advancement facilitates a deeper mechanistic understanding of metastatic spread.
• Enables large-scale simulations crucial for developing new cancer therapies and diagnostic strategies.