Tool for designing and running computational experiments in systems biology.
- Experiment designer provides a way to specify features, levels, and experiments (specifications of features and levels to evaluate).
- Experiment runner executes one or more previously specified experiments and collections statistics. Experiments can be run in parallel using
dask. - Experiment planner provides time estimates for running a set of experiments.
- A factor is something that is changed in an experiment, such as start concentration of a species and algorithm for parameter estimation.
- A level for a factor is a value that is assigned the factor for an experiment.
- Factor space has dimensions of factors, and the coordinates for a factor are its levels.
- A condition is a collection of pairings of factors with their level so that all factors are assigned a level.
- An experiment is the execution of a condition that produces a collection of results.
- A work unit is a collection of experiments. Typically, the conditions for experiments are chosen as a hypercube in factor space.
The user specifies (or manually creates) Factor objects that have attributes of name and data type for their level.
A Condition is a dictionary of Factor keys whose value is the level for the Factor.
An Experiment is a condition paired with an ExeprimentExecution, a function that can interpret a condition,
run an experiment, and produce an ExperimentResult.
A Workunit is a collection of Experiments that are submitted to the ExperimentRunner, which provides capabilities for monitoring progress and halting execution.
ExperimentScheduler returns 0 or more ExperimentResult.