Lengthening the timescale reach of molecular dynamics

molecular dynamics;implicit solvent;Graphics Processing Units (GPU);High Performance Computing (HPC);flagellum;protein translocation

Tanner, David

关键词: molecular dynamics;    implicit solvent;    Graphics Processing Units (GPU);    High Performance Computing (HPC);    flagellum;    protein translocation;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/31144/Tanner_David.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Molecular dynamics (MD) is a computational method employed for studying the dynamics of nanoscale biological systems on nanosecond to microsecond timescales. Using MD, researchers can utilize experimental data from crystallography and cryo-electron microscopy to explore the functional dynamics of biological systems. The timescale reach of the molecular dynamics tool is limited by how fast femtosecond timesteps can be sequentially integrated; today's fast computers allow simulation speeds of tens of nanoseconds of simulation time per day, which typically limits simulation lengths to hundreds of nanoseconds. This work explores three ways whereby the timescale reach of molecular dynamics can be lengthened beyond nanoseconds, to the millisecond timescales of cellular processes. First, a theoretical model of flagellin translocation allows nanosecond timescale MD simulations to explore the hour-long process of bacterial flagellum elongation. Second, a generalized Born model of implicit solvent accelerates simulation through reduced computational expense as well as increased conformational sampling due to reduced viscosity of the implicit solvent. Finally, advanced computing technologies, such as graphics processing units, accelerate simulation speeds of hybrid GB/SA implicit solvent models, thereby directly increasing simulation lengths.