En route for molecular dynamics simulation of a living cell

Abstract

Creating an in silico all-atom whole-cell model for molecular dynamics (MD) simulation is one of the best ways to quantitatively understand the basic structure and function of cells in terms of the laws of physics and chemistry. The heavy use of graphics processing units (GPUs), the exponential growth of supercomputing power, and the emergence of MD simulation-specific supercomputers lay the groundwork for the MD simulation of molecular machinery. Moreover, the involvement of artificial intelligence (AI) will not only improve the accuracy of the simulation but also significantly accelerate the sampling efficiency. However, several underlying critical puzzles prevent in silico all-atom whole-cell modeling, which is the holy grail of MD simulation. From this perspective, we briefly reviewed the accomplishments of present techniques and hardware as well as provided insight to address the challenge of MD simulation of a living cell. With the rapid advancements in computational hardware, AI, and experimental cell biology, it would be possible to achieve this overarching goal.