Parameter Estimation

Parameter Estimation (both in terms of allowing the user to search for good pameter values to a model by hand, and allowing the computer to perform automated searches for good pameter values) is a major focus of the JigCell project. We have done extensive work toward building tools to support both user-based and automated parameter estimation. However, at this time the parameter estimation tools are not integrated with the other components of the JigCell PSE.

A parameter estimator takes a biological model, experimental data, and the relationship between the model and data. Using this information the parameter estimator uses numerical tools to vary the parameters to the model looking for the parameters that best fit the experimental data.

Our most current parameter estimation software is PET (Parameter Estimation Toolkit).

If you wish to check out our earlier parameter estimation tools, we still maintain access to BioPack. The user of BioPack must give the estimator a biological model in the form of ordinary differential equations, experimental data in the form of numbers, and the relationship between the model and data in the form of FORTRAN 90 code. The first two of these is fairly trivial for someone already involved in biological models. The last can be a difficult task for anyone not familiar with both biological modelling and FORTRAN 90. The eventual goal is to make it as easy as possible to enter all of the necessary information into JigCell.

BioPack depends on ODRPACK for minimizing the error between the model data and experimental data. The parameter estimator depends on LSODAR to solve the system of ODE's numerically. The solutions are used to generate model data which is fed into ODRPACK as information used to adjust the parameters. Here is a system diagram showing how everything fits together inside the parameter estimator.

The following is a catalog of links to the earlier work. However, we encourage users to try out PET.