Model reduction

Description

The reduction of regulatory graphs allows to extract a "simplified" regulatory graph where a set of components are hidden. To keep a consistent dynamical behaviour, the logical rules associated with the targets of each hidden component account for the (indirect) effects of its regulators. This construction of reduced models preserves crucial dynamical properties of the original model, including stable states and more complex attractors. Furthermore, the relationship between the attractor configuration of the original model and those of reduced models is formally established. Further information on the algorithm is available in Naldi et al. (TCS 2011).

Output stripping

Outputs are components which do not regulate others. As such, these components have no impact on the attractors that will be reached in a simulation. These output components can be automatically removed when performing a simulation or some other actions on a model. Output reduction was introduced in Naldi et al. (CMSB 2012).

Propagate fixed components

The components with a fixed target value can be reduced without affecting the attractors (but their reachability can be compromised depending on the initial state). This reduction is extended automatically to their targets which may become fixed as well.

We plan to extend this to input components which are fixed by the initial conditions.

Eliminate duplicate components

If several components share the same logical function, only one is kept, and the others are reduced to mirrors of this reference component. This automated reduction can affect cyclical attractors and rechability properties.

Usage

This modifier requires some configuration. The supported perturbations and the syntax of the parameter string to use in command line or scripting mode are described below. We remind that in command line, the parameter string for model reduction is given in the -m reduction:parameter switch. In scripts, it is the last argument in lqm.modifyModel(model, "reduction", parameter).

This modifier takes as parameter a comma-separated list of keywords to select the performed reduction. fixed, output, and duplicate respectively enable the propagation of fixed components, output stripping, and elimination of duplicates. fixed,output enables the two selected actions.

Future plans include support for additional reduction strategies and the specification of custom lists of reduced components.

Going further

The ReductionService, instance will provide additional methods to facilitate the definition of custom reduction strategies.