Lead PI:
Center Researchers:
The Challenge
Understanding how cells interact within the human bloodstream is crucial for advancing biomedical research and developing new treatments. The circulatory system is a dynamic environment where cells constantly interact with one another and their surroundings. Tracking individual cells—such as cancer cells, immune cells, or drug-delivery particles—requires capturing complex biomechanical forces and fluid dynamics across vastly different spatial scales.
One of the most pressing applications of this research is metastasis, the process by which cancer spreads to new parts of the body. Metastasis is responsible for over 90% of cancer-related deaths, yet predicting where circulating tumor cells (CTCs) will arrest and form secondary tumors remains an open question. Physical interactions between CTCs and red blood cells (RBCs) play a major role in determining where cancer cells become trapped, but traditional simulation methods either oversimplify these dynamics or require prohibitive computational resources.
The main challenge lies in accurately modeling the movement and interaction of individual cells at submicron resolution while simultaneously simulating their transport through blood vessels spanning centimeters to meters in length. Existing computational techniques struggle to bridge these scales, making it difficult to study critical biological processes, such as how CTCs navigate blood flow, interact with deformable RBCs, and eventually adhere to vessel walls. To address these limitations, we developed an innovative solution that drastically improves the ability to model cell behavior at biologically relevant scales.
Our Solution
To overcome this challenge, our team developed the Adaptive Physics Refinement (APR) method, a groundbreaking technique that enables detailed simulations of cellular interactions within the bloodstream at unprecedented length scales while using significantly less computational power. Using Summit at ORNL—previously the world’s largest supercomputer—conventional brute-force approaches were limited to modeling just a few cubic microns at full resolution. In contrast, APR has enabled us to simulate cancer cell transport across centimeters while maintaining cellular-level detail, achieving this breakthrough not on a massive supercomputer, but on a single cloud-based node, demonstrating an unprecedented leap in computational efficiency and accessibility.
APR introduces a dynamic, high-resolution “window” that tracks key areas of interest—such as a cancer cell traveling through blood vessels—while integrating seamlessly with a coarser, computationally efficient simulation of the surrounding circulatory system. This allows for submicron-scale modeling of cell interactions without the need to explicitly resolve the entire vascular network at such high detail.
By employing this adaptive, multi-resolution approach, APR efficiently manages different fluid viscosities and flow dynamics, making it possible to conduct detailed long-distance simulations without the need for excessive computational resources. For example, APR successfully simulated the movement of a cancer cell over a centimeter using a single node on Amazon Web Services (AWS) for 500 hours. This result represents a radical increase in computational efficiency, reducing memory requirements from petabytes to gigabytes.
The impact of APR extends beyond metastasis research. It provides a scalable framework for studying a wide range of biomedical questions, from how immune cells respond to infections to how drug-delivery nanoparticles circulate and interact with tissues. The ability to simulate cellular transport in realistic vascular environments with dynamically maintained RBC distributions opens the door to new insights into cancer progression, improved predictive models, and ultimately, better strategies for early intervention and treatment.
Future developments will further enhance APR’s capabilities by incorporating explicit modeling of cell adhesion, interactions near blood vessel walls, and immune system engagement. These advancements will continue to push the boundaries of computational biomedical research, providing a transformative tool for understanding disease progression and optimizing therapeutic interventions.