POptimizer

Description

The poptimizer is developed for optimizing both the structure and the parameters of system biological models automatically by using evolutionary algorithms.

Input Data

Currently the POptimizer can process two types of input data from the cell system biological data with single compartment:

1) the time series data of multiple target objects under one initial state.
2) the time series data of multiple target objects under different initial states.

Models

The models built by the POptimizer can have flexible model structure and parameters. The basic structure for each model takes the form of set of modules. Those elementary modules are instantiated or copied from the model library. The user can define their own non-fixed module library or fixed module library based on the field knowledge where the former means that all the modules in this library act as the 'building blocks' to combine into various modules during the evolution which the latter means that all the modules in this library are attached into each model with fix structures and parameters.

In addition the structure of each module in non-fixed module library can be partially fixed by fixing some rules and all the parameters within the module can be fixed or non-fixed by defining their specific range (the parameter is fixed if the lower bound and the upper bound are the same otherwise non-fixed). The parameter can have the linear scale value or logarithmic scale value.

Model Structure Optimization

The optimization of the model structure concerns with the choice of modules that should compose the model. The number of modules and their corresponding instantiation is also explored to minimize between the output data generated by the model and the target data. The modules that can used are specified in a library.

Model Parameters Optimization

The optimization of the model parameters concerns with learning the appropriate kinetic constants corresponding to each one of the rules specified in the modules. The module library specifies the parameter ranges (as well as the linear/logarithmic scale) for these kinetic constants.