Skip to content

Support for PEtab

Lukas Refisch edited this page Apr 28, 2020 · 3 revisions

About PEtab

The SBML model definition is an established standard in the systems biology community for communicating information about models. A full description of the parameter estimation problem is not covered by SBML.

To make parameter optimization problems interchangeable between different toolboxes and programming languages, the format PEtab (Parameter Estimation Tabulars, Publication). In general, the format consists of (exactly) one SBML-file describing the model, that is accompanied by several tab separated files (*.tsv) which describe the observation functions, the measurement values, condition specific settings, and the parameter settings for optimization. The standards of this data format are specified in full detail here.

Importing PEtab files

Import of a parameter estimation problem specified in PEtab is done via the function arImportPEtab. This replaces the calls of arLoadModel, arLoadData, arCompileAll contrary to the usual D2D workflow. The steps performed by arImportPEtab are the following:

  • Read in the model.sbml file via arParseSBML and write a model.def file
  • Call arLoadModel('model.def')
  • Construct adequate data structs that are appended to the ar.model-struct via arCreateDataStruct and arAddDataStruct
  • Account for condition specific changes via arLoadCondPEtab
  • Call arCompileAll to compile model and data
  • Call arLoadParsPEtab to load correct parameter values, bounds, etc..
  • perform preEquilibration (via arSteadyState) if specified

A script that loads the set of PEtab files my_model.xml, my_observables.tsv, my_measurements.tsv, my_conditions.tsv, my_parameters.tsv may look like this:

arInit
arImportPEtab({'my_model','my_observables','my_measurements','my_conditions','my_parameters'}) % note the order of input arguments!

Tests

In the folder Examples/ToyModels/PEtab_Tests some small test scenarios describing parameter optimization problems are defined. The script TestFeature.m will perform the tests and checks for agreement with provided values of likelihood and chi2 as well as some simulation results. The test suite was proposed in the publication to resemble a sequence of minimal examples to check the support of the PETab format.

Exporting PEtab files

Export functionality is currently in beta state.

For standard modeling projects, the goal is to automatically export your d2d modeling project into PEtab using arExportPEtab. In this function, all necessary files are written and saved to a single folder. This allows for compatibility and interoparability with other modeling frameworks in a straightforward manner.

Clone this wiki locally