[ugrads] Computational science talk
hickernell at iit.edu
Wed May 16 18:40:33 CDT 2018
Join us tomorrow for …
Computational Challenges in Multiscale Modeling of Muscle Contraction
Srboljub M Mijailovich, the Department of Biology at Illinois Institute of Technology
Thursday, May 17, 11 AM
Pritzker Science Center 240
Molecular models of contractility in striated muscle require an integrated description of the action of myosin motors, firstly in the filament lattice over the whole fiber, then over muscle tissue and toward whole organ. General multiscale model of muscle contraction includes two inter-connected models: (1) MUSICO Platform bridging the scales from inter-atomic (~ A) to whole cell (~100-500 mm) and (2) MUSICO Multiscle FEA and Diffusion Spectrum Imaging (DSI) bridging the meso-scales from molecular interactions (~ nm) to tissue (~ mm) or organ scale (tens of cm).
An integrated model up to muscle cell scale is defined by applying knowledge of the actin-myosin-ATP cycle subject to the constraint that: (a) strain-dependent actin-myosin kinetics are derived from reaction-energy landscapes which are applied to dimeric myosin, (b) actin-myosin interactions in the half-sarcomere and whole-sarcomere are defined in the context of 3D sarcomere geometry with discretely-positioned heads on the myosin filament and target zones on actin filaments, and (c) the myosin and actin filaments are treated as elastically extensible. We specifically develop two kinds of sliding filament models: (1) a 3-D comprehensive molecular model of a sarcomere contraction which includes explicit position of each myosin head and corresponding site(s) and its association with thin filament regulatory proteins; (2) probabilistic models which include most of the essential features of the 3-D molecular model and are computationally more feasible for multiscale formulations. The development of these models is supported by molecular dynamic (MD) simulations and molecular experiments. The mathematical description includes Monte Carlo simulations at molecular scale coupled with mechanical analysis with Finite Elements. The novel approach provides a method for studying the effect of contractile protein mutations on mechanochemistry, mechanotransduction, and organ dysfunction and thus will constitute a tool to study physiologically relevant disease models.
MUSICO meso-scale models, coupling FEA and DSI, are designed for simulations of a heartbeat in whole heart as a multi-scale problem. For this challenging and complex approach, it is necessary to prescribe both, electrical and mechanical characteristics of the myocardium and it requires three principal components: (1) A multiscale model coupling molecular kinetics and its regulations with local material characters for macroscale continuous mechanics (finite element analysis), (2) Mesoscale structural organization of the heart tissue and (3) Excitation-contraction coupling that is an essential step of electro-mechanical cardiac simulations. The computational platform includes nonlinear strain (displacement) based FE solvers where stress tensor in each integration point is calculated as a function of strains in the point, according to material model of the element. However, during muscle contraction, the material characteristic of muscle depends on local rate and history of deformation and cannot be measured and prescribed in integration points. Thus using a multiscale formulation the instantaneous muscle material characteristics are calculated from probabilistic models prescribed by a few kinetics and structural parameters. This can be achieved by coupling MUSICO and FE solver, where for prescribed rate of deformation, i.e. shortening velocity of muscle fiber, MUSICO can calculate in any instant the local active stress (in direction of muscle fiber) and fiber stiffness. For known instantaneous muscle active stress and stiffness, along with passive material characteristics and current boundary conditions, FE solvers will provide updated muscle deformation and rate of deformation.
The modular structure of the platform readily accommodates extension and replacement of any part of the sarcomere or higher length sale system. All of these systems are adaptable (for different muscle systems), scalable and parallelizable to address larger and larger ensembles as computing power permits. The current form includes the concept of distributed calculations of multi-scale domains in homogeneous computing environment (CPUs only) or in a mixed CPU–GPU environment designed for computationally effective simulations on variety of hardware platforms.
Fred J. Hickernell
Director, Center for Interdisciplinary Scientific Computation (CISC)
Professor, Department of Applied Mathematics
Illinois Institute of Technology
Pritzker Science Ctr Rm 106, 3105 S Dearborn St, Chicago, IL 60616
hickernell at iit.edu <mailto:hickernell at iit.edu>, www.iit.edu/~hickernell <http://www.iit.edu/~hickernell>
Office: +1 312 567 8983 Cell/WhatsApp: +1 630 696 8124 Skype/WeChat: fjhickernell
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the ugrads