Self-consistent coupling of molecular theory of solvation with - PowerPoint PPT Presentation

Self-consistent coupling of molecular theory of solvation with molecular simulations Presented by: CAO Siqin, Jan 7

Solvate with 3D-RISM-KH Bridge between solvation theory and other methods: = + Solvate with 3D-RISM-KH: Andriy Kovalenko and Sergey Gusarov, Phys. Chem. Chem. Phys. 20 , 2947 (2018)

MTS-MD/OIN/GSFE/3D-RISM-KH • MTS: multi-timescale • MD: molecular dynamics • OIN: optimized isokinetic Nosé–Hoover • GSFE: generalized solvation force extrapolation • 3D-RISM-KH: 3D reference interaction site model complemented with the Kovalenko-Hirata closure Kinetics Total force Chain-thermostat Andriy Kovalenko and Sergey Gusarov, Phys. Chem. Chem. Phys. 20 , 2947 (2018)

MTS-MD/OIN/GSFE/3D-RISM-KH: GSFE Solvation force: Solvation force extrapolation, the GSFE (general solvation force extrapolation): Coupling Scheme Translation Perform 3D-RISM every h Extrapolate force every Δ t Vacuum kinetics otherwise Andriy Kovalenko and Sergey Gusarov, Phys. Chem. Chem. Phys. 20 , 2947 (2018)

How does it save time h MTS-MD Speedup -- 1 ns/day -- 8 fs 0.15 ns/day 15% 24 fs 0.44 ns/day 44% 400 fs speedup grows slowly 1 ps 10 ns/day 10x 2~4 ps speedup saturated * Fitting of speedups: speedup = 0.1366 + 0.00986 × h /(1 fs) * 3D-RISM: 1 step RISM ≈ 51 steps explicit solvent MD Andriy Kovalenko and Sergey Gusarov, Phys. Chem. Chem. Phys. 20 , 2947 (2018)

Problem with the implicit-solvent MD Explicit Solvent MD Implicit Solvent MD Solvent Configuration Solvent Configuration Protein Conformation Protein Conformation Folding time: miniprotein 1L2Y protein G 60 ns (MD) vs 4-9 us (Experiment) Thermodynamics could be correct Kinetics could be wrong [1] my understanding [2] Andriy Kovalenko and Sergey Gusarov, Phys. Chem. Chem. Phys. 20 , 2947 (2018)

Alternatively, if only cares about thermodynamics ... Alanine dipeptide in vacuum solvated with 3D-RISM Hiroshi Iwasaki, Shinnosuke Gyoubu, Tsutomu Kawatsu and Shinichi Miura, Molecular Simulation 41 , 1015-1020 (2014), Molecular Simulation 43 , 997-1003 (2017)

Summary • Implicit solvent MD with 3D-RISM: • 3D-RISM is too slow, so extrapolation is required • Conformation of proteins can be correct • Kinetics is incorrect • An alternative scheme: solvate the vacuum free energy landscape with 3D-RISM A. Kovalenko and S. Gusarov, Phys. Chem. Chem. Phys. 20 , 2947 (2018); H. Iwasaki, S. Gyoubu, T. Kawatsu and S. Miura, Molecular Simulation 41 , 1015-1020 (2014), Molecular Simulation 43 , 997-1003 (2017)