Why Discrete Stochastic Simulation - PowerPoint PPT Presentation

1 / 5
About This Presentation
Title:

Why Discrete Stochastic Simulation

Description:

Polarization in yeast mating T. M. Yi, UC Irvine ... We need a collection of community models! Why? Facilitate development and extension of models ... – PowerPoint PPT presentation

Number of Views:34
Avg rating:3.0/5.0
Slides: 6
Provided by: sabou
Category:

less

Transcript and Presenter's Notes

Title: Why Discrete Stochastic Simulation


1
Multiscale Simulation for Biochemical Systems
  • Why Discrete Stochastic Simulation?
  • An ODE model cannot capture effects due to small
    numbers of key chemical species
  • A molecular dynamics model is too slow given the
    model complexities, time scales of interest, and
    need for repeated runs
  • Why Multiscale?
  • A discrete stochastic model of all species and
    all reactions may be too slow. Accelerate
    and/or couple to ODE model
  • Parameters could be coming from finer scale (MD)
    simulations

2
Software What do we have?
  • Discrete stochastic simulation of well-mixed
    systems MCell, StochKit, Cain/SSALIB, JigCell,
    CellDesigner, CellMC, STOCKS, StochSim, SynBioSS,
    TinkerCell, Moleculizer,
  • Some limited multiscale capabilities

3
Software What do we need?
  • Stiff stochastic systems
  • Adaptive multiscale
  • Probabilities of rare events
  • Sensitivity analysis, uncertainty quantification,
    reduced order stochastic models
  • Parameter estimation
  • High performance, for example GPU

4
Spatial stochastic simulation
  • Huge problem of computational complexity
  • Large numbers of diffusive transfers must be
    aggregated to enable parallelism
  • Intrinsic physical limitations put lower bound on
    spatial grid size
  • Mathematical and software infrastructure must
    support complicated geometries, unstructured
    meshes
  • Molecular crowding effects reaction propensities,
    but by how much?

Unfolded protein response in the endoplasmic
reticulum C. Young, A. Robinson, U. Delaware
Polarization in yeast mating T. M. Yi, UC Irvine
5
We need a collection of community models!
  • Why?
  • Facilitate development and extension of models
  • Experiment with published models and
    validate/invalidate with new experimental data
  • Demonstrate the issues that computational
    infrastructure need to address
  • Test the viability and efficiency of new
    algorithms and software on realistic models
  • Issues
  • Requires a common modeling language and/or
    standards and expectations for publication
  • The language must be extensible to accommodate
    unforseen future needs
Write a Comment
User Comments (0)
About PowerShow.com