Simulation Programs: What is out there? A critical evaluation. Prof:Rui Alves email@example.com 973702406 Dept Ciencies Mediques Basiques, 1st Floor, Room 1.08 Website of the Course:http://web.udl.es/usuaris/pg193845/Courses/Bioinformatics_2007/ Course: http://10.100.14.36/Student_Server/
Simulation is becoming widespread • Kinetic models are becoming a common tool for testing biological hypothesis. • A plethora of different software packages for model building, simulation and analysis is available. • How far are we from having reliable tools that make simulation accessible to non-expert mathematical modelers? • What tools are more adequate for different types of problems?
Addressing the questions • Choose representative simulation packages. • Identify features in each of them. • Test how accurately these features work. • Evaluate how much expertise one needs to use each program.
Emphasis • Simulators • Type of input • Models used for testing • Analytical capabilities of the software • Cross-compatibility between the software • What to use for each type of problem
Rationale for excluding software • Kinetic Modeling Packages that require expert knowledge about the mother-software excluded • Not for non-experts + expensive • Commercial user-friendly simulation software provides no new functionality with respect to free software. Also, we could not get temporary evaluation licenses for some of them, so we excluded them all.
Evaluated Simulation Packages • Kinetic Modeling Packages • Excludes software that implement functionality in a pre-existing software platform (e.g. BST lab in MATLAB). • Free Stand alone simulators • Excludes e.g. STELLA and MADONNA • Free Internet Simulation Servers • Free Network editors
Rationale for choices of models • Models that allowed the testing of the different features. • In some cases model with analytical solutions so that accuracy of calculations could be determined
Models used in the evaluation (1) • Escherichia coli’s phosphoenolpyruvate:glucose phosphotransferase system (mass action; 1 compartment) • GAL4 system of Saccharomyces cerevisiae (mass action; 1 compartment) Tests for stochastic simulators
Models used in the evaluation (2) Source reactions Reactions with modifiers Reactions with catalysts Homo molecular reactions
Models used in the evaluation (3) • Simple difusion two compartment model
Models used in the evaluation (4) • Simple two reaction model with analytical solution to evaluate sensitivity and stability analysis X1 X2 Tests for stability and sensitivity analysis as well as moiety conservation calculations
Analytical capabilities of the software Incorrect for models with moiety conservation Incorrectly implemented jacobian calculations
Compartmental model implementation model 4 Write all equations/Draw diagrams Choose Kinetic equations/Write KEs Interface reaction parameters must be correct to have the appropriate units Permeability constant converted into apparent rate constants Kinetic parameters multiplied by volume of corresponding compartment
Exceptions: V-Cell Dizzy – Convert everything into ammounts rather than concentrations Cellware – Chose a reference compartment and convert all concentrations to that compartment
Crosstalk between software • It is important to be able to share models between different software programs • SBML is becoming the standard
Trouble in SBMLland • reaction stoichiometries defined as floating point values; • boundary metabolites; • source reactions • sink reactions • reactions where the stoichiometry for one of reactants or products is larger than 1; • kinetic type definitions can prevent correct interpretation of models by stochastic simulators. • Definition of compartments breaks down for variable volumes
There is hope in SBMLand • A small amount of editing is in general sufficient to correct imcompatible SBML models • A redefinition of comparments is straightforward
Goals of the model (I) • Large scale modeling • Reconstructing the full network of the genome • Red Blood Cell Metabolism • Dialog or diagram based, with possibility for modular implementation • COPASI, GEPASI, V-CELL, CELLDESIGNER,JDESIGNER • Sensitivity analysis • COPASI, GEPASI, jdesigner, v-cell
Goals of the model (I) • Modeling Specific Pathways/Circuits • Non-catalytic lipid peroxidation • MAPK Pathways • Any type of input; • Sensitivity analysis • COPASI, GEPASI, jdesigner, PLAS, v-cell, celldesigner
Goals of the model (I) • Generating alternative hypothesys for the topology of the model. • Must allow for structured functional forms and for large scale parameter scans • COPASI, GEPASI, jdesigner, PLAS, celldesigner
Goals of the model (II) • Estimating parameter values • Must have fitting algorithms • COPASI, GEPASI,DYNAFIT • Identifying Design Principles • None, so far; howver we have a MATHEMATICA package that allows you to do this.
Final Conclusion • Our analysis has convinced us that a non-trivial degree of expertise is still required for the use of simulation programs to create models. • It is dangerous to expect that a non-expert will create a useful and correct model of a biological process. Alves et al. Nature Biotechnology 2006