leipzig, 17 may 20071 markov models of protein folding - application to molecular dynamics...
Post on 22-Dec-2015
215 views
TRANSCRIPT
Leipzig, 17 May 2007 1
Markov Models of Protein Folding
- Application to Molecular Dynamics Simulations
Christian Hedegaard Jensen
Leipzig, 17 May 2007 2
Outline
• Markov Models and Molecular Dynamics Simulations
• Investigating Accelerated Molecular Dynamics
Leipzig, 17 May 2007 3
Outline
• Markov Models and Molecular Dynamics Simulations
• Investigating Accelerated Molecular Dynamics
Leipzig, 17 May 2007 4
Basic Idea
• Do many Molecular Dynamics Simulations
• Cluster the trajectories into discrete states
• Combine the information about transitions from all the trajectories into a single Markov model transition matrix.
• This allows for highly parallel simulations. E.g. Folding@Home
Leipzig, 17 May 2007 5
How Sensitive is the Dynamics to Clustering
• Investigate a small peptide VPAL• 500ns MD simulation with explicit
water at 300k• How sensitive are the mean first
passage times between states, to the clustering of the states.
Leipzig, 17 May 2007 11
Outline
• Markov Models and Molecular Dynamics Simulations
• Investigating Accelerated Molecular Dynamics
Leipzig, 17 May 2007 12
Accelerated Molecular Dynamics
• Modify the potential energy landscape to speed up transitions between states.
D. Hamelberg, J. Mongan, and J. A. McCammon. Accelerated molecular dynamics: a promising and e±cient simulation method for biomolecules. Journal of Chemical Physics, 120(24):11919-11929, 2004.A. F. Voter. Hyperdynamics: Accelerated molecular dynamics of infrequent events. Physical Review Letters, 78(20):3908-3911, 1997.A. F. Voter. A method for accelerating the molecular dynamics simulation of infrequent events. Journal of Chemical Physics, 106(11):4665-4677, 1997.
Leipzig, 17 May 2007 13
Scaled Potentials
• Scale potential by constant
• Energy differences
• Assuming transition state theory: similar to changing temperature
10 ~ VV
2121 VVVV
TkEB
r *exp
Leipzig, 17 May 2007 14
Scaled Potentials
• Possible use 1:– Do a simulation with <1
• Possible use 2:– Start with =1– For every e.g. 10ps in same minima change =
0.9*– When in new minima reset =1
• This means that the potentials are modified just enough (and no more) to escape a minima within a reasonable time
Leipzig, 17 May 2007 15
Scaled Potentials
• Do calculations with different and investigate the change in mean first passage times between states.
• Compare to simulations done at different temperatures.
• If there is a significant speedup implement algorithm.