Multiscale Modeling of Mechanotransduction Processes in Flow-Induced Platelet Activation

Shear induced platelet activation is the major cause of thrombus formation in cardiovascular diseases and prosthetic devices. A novel multiscale modeling approach based on discrete particle methods permits the investigation of such complex phenomena by coupling the macroscopic flow conditions with the cellular and molecular effects of platelet activation. However, such a modeling approach poses a major computational challenge, where a typical simulation involves the computation and communication of millions of particles. The large computational requirement was fulfilled with the assistance of the state of the art high performance computing facilities. In this paper, we first present our multiscale model of shear-induced platelet activation, where the blood plasma and platelet were modeled by dissipative particle dynamics (DPD), and coarse grained molecular dynamics (CGMD), respectively. We then employ the blood plasma-platelet interaction model to study the platelet flipping dynamics and mechanotransduction process of platelet activation. To achieve the best performance of our model on supercomputers, we conduct the performance analysis by scaling our blood plasma model with 20 million particles to 10, 000 CPU cores on Sunway bluelight system. The simulation results suggest that our multiscale model of shear-induced platelet activation can fully achieve fluid-platelet interaction across micro-and nano-scales, and can achieve good strong scalability running on the system with large number of processors.