Scalable Molecular Dynamics

1
Scalable Molecular Dynamics

• T.P.Straatsma
• Laboratory Fellow and Associate Division Director
• Computational Biology and Bioinformatics
• Computational Sciences and Mathematics Division
• Pacific Northwest National Laboratory

2
NWChem Molecular Science Software
3
Domain Decomposition
4
Force Evaluation
5
Particle-mesh Ewald
1. Charge grid construction 2. Block to slab
decomposition 3. 3D-fast Fourier transform 4.
Reciprocal space energy forces 5. 3D-fast
Fourier transform 6. Slab to block
decomposition 7. Atomic forces
6
Flowchart
7
Timing Analysis
Haloalkane dehalogenase, force evaluation timings
8
Load Balancing
Collective Resizing
9
Dynamic Load Balancing
10
Challenges for the DOE

• Environmental Legacy at Hanford and other DOE
  sites
• Bioremediation

• Environmental and Health Impact of Energy Use
• Carbon sequestration
• Nitrogen fixation

• Production of Energy
• Biofuels
• Hydrogen

11
Molecular Basis for Microbial Adhesion and
Geochemical Surface Reactions

• Microbes in the subsurface mediate a number of
  environmental, geochemical processes
• Uptake of metal ions, including environmentally
  recalcitrant metals
• Adhesion to mineral surfaces
• Reduction and mineralization of ions at the
  microbial surface
• Pseudomonas aeruginosa Cu, Fe, Au, La, Eu, U,
  Yb, Al, Ca, Na, K
• Shewanella putrefaciens Fe, S, Mn
• Shewanella alga Fe, Cr, Co, Mn, U
• Shewanella amazonensis Fe, Mn, S
• Shewanella oneidensis MR1 External reduction
  involving OM cytochromes

12
Project Objectives

• Molecular level characterization of
• Microbial adhesion to mineral surfaces
• Metal ion concentration in microbial membranes
• Focus on Gram-negative bacterial Outer Membrane
• Computational Approach
• Molecular modeling and molecular dynamics
  simulations
• Quantum mechanical description of key functional
  groups
• Thermodynamic Modeling

13
Gram Negative Cell Walls
14
LPS of Pseudomonas aeruginosa
1. Design of the Rough LPS Molecular Model 2.
Determination of Electrostatic Model
15
LPS Membrane Construction
Distribution of functional groups and water in
the outer membrane of P. aeruginosa. These
results are used for thermodynamic modeling of
ion adsorption in microbial membranes.
16
Phosphate Clustering
Outer Core
Inner Core
These results lend support to the interpretation
of recent XAS experiments carried out by J.
Bargar at SLAC indicating that uranyl ions take
up by microbial membranes exists in clusters
involving phosphates.
17
Membrane Electrostatic Potential
Average Potential Across Membrane Calc. 100
mV Exp. 80 mV
18
Atomic Charges from 2D SCF-HF ESP Fit
19
Membrane-Mineral Interactions
20
P. Aeruginosa Outer Membrane Proteins
E. coli membrane protein FecA (Pautsch and
Schultz, 1998) and homology modeled P. aeruginosa
membrane protein FecA (Straatsma, unpublished)
E. coli membrane protein TolC (Pautsch and
Schultz, 1998) and homology modeled P. aeruginosa
membrane protein OprM (Wong et al., 2001)
E. coli membrane protein OmpA (Pautsch and
Schultz, 1998) and homology modeled P. aeruginosa
membrane protein OprF (Brinkman et al., 2000)
21
P. aeruginosa OprF
22
Electron transfer in bacterial respiration

• Under anaerobic conditions, Shewanella
  frigidimarina is able to use extra-cellular iron
  as the electron acceptor in its respiration.
  The electron transfer pathway involves a number
  of cytochromes which deliver electrons from the
  cytoplasmic membrane to the periplasmic membrane,
  where iron reduction occurs.
• The electron transfer (ET) between the membranes
  is carried out by the respiratory enzyme
  flavocytochrome c3 fumarate reductase (Fcc3),
  which contains four bis(histidine) hemes.

23
Electron Transfer in Fcc3 and Ifc3
Flavocytochrome c3 fumarate reductase of
Shewanella frigidimarina
24
Marcus theory of electron transfer
25
B3LYP Characterization of a model heme
ET donor/acceptor orbital dp
26
Computational Structural Biology Challenges

• Computational protein structure prediction
• Protein-protein complexes cell signaling
• Protein-membrane and mineral-membrane complexes
• Extension to microsecond simulation times
• Statistically accurate thermodynamic properties
• Comparative trajectory analysis
• Enzyme catalysis using hybrid QM/MM methods
• Extension toward millisecond simulation times
• Protein folding and unfolding
• Membrane transport of simple ions and small
  molecules
• Membrane fusion, vesicle formation
• Scalability on next generation MPP and hybrid
  architectures

27
Acknowledgements

• Dr. Roberto D. Lins, ETH Lausanne, CH
• Dr. Robert M. Shroll, Spectral Sciences, Boston,
  MA
• Dr. Wlodek K. Apostoluk, Wroclaw University,
  Poland
• Dr. Andy R. Felmy, Chemical Sciences Division,
  PNNL
• Dr. Kevin M. Rosso, Chemical Sciences Division,
  PNNL
• Professor David A. Dixon, University of Alabama
• Dr. Erich R. Vorpagel, EMSL
• DOE Office of Advanced Scientific Computing
  Research
• DOE Office of Basic Energy Science, Geosciences
  Research Program
• DOE Office of Biological and Environmental
  Research
• EMSL Molecular Sciences Computing Facility