2023年度第2回計算科学フォーラム

Cheng Tan

（理化学研究所R-CCS 粒子系生物物理研究チーム 特別研究員／兵庫県立大学 情報科学研究科 准教授）

「Developments and Applications in Multiscale Molecular Dynamics Simulations for Biomolecular Condensation Using GENESIS」

Biomolecular condensates, formed through liquid-liquid phase separation (LLPS), play a pivotal role in numerous biological processes. However, their underlying mechanisms remain largely unexplored due to experimental constraints and the computational challenges posed by the vast number of particles involved. To overcome these obstacles, we developed multiscale computational models in the molecular dynamics (MD) software GENESIS, significantly enhancing its performance on supercomputers like Fugaku. In this study, we present large-scale MD simulations of condensates involving the protein TDP-43 and its regulatory counterpart, Hero11. Initially, we employed coarse-grained (CG) simulations, in conjunction with the slab method, to model the phase behavior of homotypic TDP-43 condensates and heterotypic condensates comprising both TDP-43 and Hero11. We then expanded our system to include approximately 16,000 proteins, approaching the length scale of hundreds of nanometers. These large-scale CG simulations, carried out on Fugaku, enabled us to monitor the dynamics of droplets over an extended time scale (10^9 time steps). Subsequently, we used our CG structures to reconstruct all-atom models of the protein condensates. The reconstructed systems, comprising approximately 2.5 million atoms, were simulated for around 3 microseconds. Our analysis focused on protein secondary structures, atom distributions, and the roles of water molecules and ions within these systems. Our findings offer a more profound insight into the factors that affect the stability and morphology of protein condensates. Furthermore, the large-scale MD simulations we conducted have proven to be an invaluable tool in deciphering the complex mechanisms behind biological phenomena.

発表資料（PDF:2.05MB）

吉川 耕司

（筑波大学計算科学研究センター 准教授）

「重力多体系のVlasovシミュレーションとその先」

宇宙物理学・天体物理学では重力相互作用は天体形成にとって本質的な物理過程であり、これまでの天体形成における数値シミュレーションでは重力N体シミュレーションが用いられてきた。本講演では、N体シミュレーションに代わる手法として、位相空間上の分布関数が従う無衝突Boltzmann方程式（Vlasov方程式）を直接数値的に解くVlasovシミュレーションとその宇宙論的ニュートリノへの応用やN体シミュレーションとの比較について紹介する。また、Vlasovシミュレーションの他分野への応用の可能性や将来のスーパーコンピュータを見据えた展望について述べたい。

発表資料（PDF:4.91MB）