PEtab control: a PEtab extension for optimal control problems.

Installation

Clone this repository then install it into your Python (virtual) environment. Currently, you also need to install PEtab Timecourse.

git clone https://github.com/dilpath/petab_control
cd petab_control
pip install -e .

Notes

• the time of objective measurements and controls are expected to be relative to the start_time of the control problem
  • i.e., for 20 time units after the last measurement timepoint of the original PEtab problem
    • set 20 as the time in the objective measurements or controls table
    • set last_measured_timepoint as the start_time in the PEtab Control YAML file

File format

Control parameters

This is a PEtab parameters table, for the parameters that act as controls in the model.

Controls

A TSV file that defines the times at which the control parameters can take different values.

control_id	parameterId	time	value
(Unique) [string]	[string]	[float] OR [; delimited list of float]	[string/float]

• control_id: An ID for the control.
• parameterId: The control parameter.
• time: The control parameter takes the control value at these time(s).
• value: The control value, either fixed to some float value, or specified to be estimated with estimate. If estimate, this value will be estimated according to the control parameter, but independently of other controls that use the same control parameter.

Objective measurements

A PEtab measurements table. The times and values are the target for the optimal control problem. e.g., if you want to see a value of "50" at time "25", create such a measurement here.

Objective observables

A PEtab observables table, for the corresponding objective measurements. The noise model can be used to priotize targets.

PEtab Control YAML file

Groups all files together.

problems:
  - problem_id: ...
    start_time: ...
    timecourseId: ...
    control:
      control_files:
        - ...
      control_parameter_files:
        - ...
    objective:
      measurement_files:
        - ...
      observable_files:
        - ...

The YAML contains a list of optimal control problems. Each problem has:

• problem_id: An ID for the optimal control problem.
• start_time: An offset for all times in the optimal control problem.
• timecourseId: The timecourse that will be controlled.
• control_files: The "Controls" TSV files.
• control_parameter_files: The "Control parameters" TSV files.
• measurement_files: The "Objective measurements" TSV files.
• observable_files: The "Objective observables" TSV files.

TODO

• if there is an inf time point in the
• timecourses with multiple equilibrations are possible
  • if a measurement has multiple equilibrations, all measurement times should be noted with the following syntax
    • {EQUILIBRATION_INDEX}:{TIME} instead of {TIME}
    • alternative: {PERIOD_INDEX}:{TIME RELATIVE TO PERIOD OR LAST EQUILIBRATION?}