While biosystems have been analyzed using models appropriate for phenomena on given spatio-temperal scales, efforts to combine them into a unified, multiscale model have been hampered due to ambiguities arising from the need to interface the models. The theme of this minisymposium is that a rational development of the muliscale model should start with the finest scale model and then through various mathematical techniques the coarse-grained models should follow. Techniques of interest include homogenization and renormalization group analysis, information theory and statistical ensembles, asymptotic expansions, and multiscale computational approaches. The finest scale equations providing the starting points for the analysis include molecular physics (classical and quantum mechanical) and hydrodynamics. The benefits to the pure and applied life sciences that will follow this deductive approach will be models that require a minimum of calibration (e.g., the interatomic force field). Systems of interest will range from the nanoscale (e.g., viruses, nanocapsules for drug dilivery, and intracellular structures) to whole complex organisms.

Selection of challenges, presentations from MSM folks that would highlight science and key areas where novel mathematical models might help, international supercomputing projects and computing

Back to Overall SIAM Schedule

Minisymposium_planning

Abstracts

Coarse-Grained Molecular Models of Protein Complexation -- Enghui Yap and Teresa Head-Gordon A fundamental mathematical challenge relevant to biomolecular simulation is the quantitative description of protein complexation to derive molecular machines that control biological processes such as cell signaling. We have recently achieved a fundamental result in deriving an analytical solution for computing the screened electrostatic interaction between arbitrary numbers of proteins of arbitrarily complex charge distributions, assuming they are well described by spherical low dielectric cavities in a higher dielectric salty medium [1]. By exploiting multipole expansion theory for the screened Coulomb potential, we can now describe direct charge-charge interactions and all higher-order cavity polarization correctly at all separation distances. Recently we have been extending this electrostatic model to new numerical formulations that better describe realistic shape of the protein molecules that span solutions from low to high spatial resolutions. While alternative numerical approaches are faced with an inherent trade-off between spatial resolution and memory and time requirements, which limits them to system of few macromolecules, we believe that our method is efficient and fast to compute even for many macromolecules, including frequent updates of changes in their charge distributions due to induced conformational changes. Ultimately, smooth and systematic increase in spatial resolution back to greater molecular detail of the dielectric boundaries will be the centerpiece of a multiscale scheme that converges on an atom centered solution [2]. We believe this will realize wide application in modeling complex materials problems including spatial organization of protein complexes.

[1] I. Lotan & T. Head-Gordon (2006). An analytical electrostatic model for salt screened interactions between multiple proteins J. Comp. Theo. Chem. 2, 541-555.

[2] E.-H. Yap, N. Lux Fawzi & T. Head-Gordon (2007). A coarse-grained a-carbon protein model with anisotropic hydrogen-bonding. Proteins, Struct. Func.. Bioinf. 70, 626-638.

Computational Multi-Scale Modeling in Protein-Ligand Docking -- Michela Taufer

In our multi-scale approach to protein-ligand docking, increasingly sophisticated docking models can be constructed along three scales of docking assumptions: (1) representation of the protein and ligand (potential energy function and flexibility of protein and ligand); (2) representation of the effect of solvent; and (3) sampling strategy. Computationally expensive docking models may be optimized for accuracy for difficult test cases, but may not improve results for simple test cases. It is our goal to increase the overall efficiency of docking by considering dynamic adaptations along these multi-scales to consider and evaluate new docking models. Using classifications of protein-ligand complexes based on the flexibility of protein and ligand, docking models can be dynamically adapted when more expensive models are required for sufficient accuracy. These classifications can be used to start an iterative search for adaptive docking models that optimize accuracy and minimize time to solution. The results of our project are being used to develop a Dynamically Adaptive Protein-Ligand Docking System (DAPLDS), which would be able to dynamically adapt to an appropriate docking model for new protein-ligand complexes or a large series of ligands. Our project Docking@Home (http://docking.cis.udel.edu/) uses the "volunteer computing" paradigm, which uses the Internet to harness the computing power and storage capacity of a volunteers computer resources owned by the general public. Given a variety of docking tasks, dynamic scheduling policies can selectively devote available resources based on computational expense to improve project throughput. File:Armen talk MSM.pdf

Analysis of Coarse-Grained Nucleic Acid Free Energy Landscapes -- Niles Pierce

We describe algorithms for analyzing the thermodynamic and kinetic properties of coarse-grained DNA and RNA free energy landscapes. The utility of these methods will be demonstrated by elucidating the empirical behavior of synthetic nucleic acid systems with metastable states.File:Pierce talk MSM.pdf