JSim Simulation System

1
JSim Simulation System
BREEZE Demonstration/Discussion of the JSim
Simulation System (open software for modeling and
data analysis) and the Physiome.org website for
model archiving. Dr. James Bassingthwaighte,
Erik Butterworth, Gary Raymond, Brian Carlson,
Kay Sterner, Eric Lawson University of
Washington Department of Bioengineering 23 Aug
2007
2
(No Transcript)
3
System Requirements

• Easy creation/modification/verification of models
• Diverse numerical methods
• High performance (multiprocessing if possible)
• Real-time graphical feedback
• Parameter space exploration
• Sensitivity analysis
• Optimization experimental data fitting
• Collaboration
• standard formats
• multiple platform support

4
Modeling Software Formats

• Modeling Systems (quicker development, limited
  scope)
• Berkeley Madonna ODEs only, graphical
  development
• Gepasi/Copasi ODEs only,
• Programming Systems (longer development, general
  scope)
• C/Fortran very general, longest devepment time
• Matlab widely used, , poor real-time (S-func)
  perf.
• xppaut ODEs PDEs
• Standardized Model Formats
• SBML widely supported, ODEs only, no units
• CellML limited current support, ODEs PDEs (but
  BCs need work), units

5
.

• Simcon (1967, Knopp, Bassingthwaighte, et al)
• Fortran models, terminal text/graphics
• Param. explore, sens. analysis, optimization
• XSIM (1992, Weissman,King,Butterworth)
• X-Windows, Unix, client-server computation
• Custom GUI, ad-hoc query exp(-t)F/V
• JSim (1999, Butterworth, Li)
• MML, Windows/Mac/Linux/WWW applets
• Standard formats, Multi-processing

6

• Software system for creating quantitative
  numerical models analyzing them with respect to
  experimental data
• Modeling Language (MML)
• diverse numeric methods (ODEs, PDEs, implicit
  equation, events, procedural code)
• MML Compiler / Run-time engine
• multiprocessing support
• Analysis Methodology
• easily editable models for what if? scenarios
• model structure and data visualization
• parameter space exploration
• sensitivity analysis
• parameter optimization for curve fitting
• Wide availability
• Windows, MacOS, Linux binary and source
  distributions available
• runs WWW applet
• commitment to open formats

7
JSim's Mathematical Modeling Language (MML)

• Mathematics
• ODE, PDE, Implicit
• Discrete events
• Procedural code
• Java C (currently)
• Unit balance
• Reusable templates

Leike A. Eur. J. Phys. 23 21-26 (2002)
8
JSim Analysis Methodology

• Run types
• Single runs
• Loops
• Sensitivity analysis
• Optimization / curve fitting
• Functional imaging (multiple optimizations)
• Project files
• store related set of models, parameter sets, data
  set, plot pages and notes
• easy restoration of entire analysis
• OS-independent XML, facilitates collaboration

9
Launch JSim Demo
10
Collaboration

• Model code
• CellML/SBML import (export in development)
• C, Simulink export via JSim plug-ins
• Numeric data
• 8 formats currently supported, more under
  development
• APIs
• insert Java/C/Fortran code into MML models
• extend JSim analysis I/O functions via plug-in
  modules
• physiome.org
• JSim binaries (Linux, MacOS, Windows)
• JSim source code documentation
• Curated model archives with runnable JSim applet
  models

11
Archiving and Sharing

• Contribute to developing databases Biomodels,
  CellML, SBML, Physiome.org, QKDB (quantitative
  kidney database). (Physiome.org will accept
  models in any reproducible form or language.)
• Start new sites focusing on particular purposes
• Foster adherence to model standards (though the
  standards on the IMAG site (WG10) need
  simplification. Balance checking.
• Encourage use of units and automated unit balance
  checking. Other validation methods check vs
  analytical solutions.
• NLM archive of models?

12
Explore Physiome.org

• www.physiome.org
• www.physiome.org/Models
• Model wikis models that can be run over the web
  or downloaded.
• (Jsim can be downloaded. Free of course.)