Temporary and Permanent Control of Partially Specified Boolean Networks

Investor logo

Warning

This publication doesn't include Faculty of Sports Studies. It includes Faculty of Informatics. Official publication website can be found on muni.cz.
Authors

BRIM Luboš PASTVA Samuel ŠAFRÁNEK David ŠMIJÁKOVÁ Eva

Year of publication 2023
Type Article in Periodical
Magazine / Source BIOSYSTEMS
MU Faculty or unit

Faculty of Informatics

Citation
Web https://www.sciencedirect.com/science/article/pii/S0303264722001769
Doi http://dx.doi.org/10.1016/j.biosystems.2022.104795
Keywords Boolean network; Perturbation; Temporary Control; Permanent Control; Symbolic Algorithm
Description Boolean networks (BNs) are a well-accepted modelling formalism in computational systems biology. Neverthe-less, modellers often cannot identify only a single BN that matches the biological reality. The typical reasons for this is insufficient knowledge or a lack of experimental data. Formally, this uncertainty can be expressed using partially specified Boolean networks (PSBNs), which encode the wide range of network candidates into a single structure. In this paper, we target the control of PSBNs. The goal of BN control is to find perturbations which guarantee stabilisation of the system in the desired state. Specifically, we consider variable perturbations (gene knock-out and over-expression) with three types of application time-window: one-step, temporary, and permanent. While the control of fully specified BNs is a thoroughly explored topic, control of PSBNs introduces additional challenges that we address in this paper. In particular, the unspecified components of the model cause a significant amount of additional state space explosion. To address this issue, we propose a fully symbolic methodology that can represent the numerous system variants in a compact form. In fully specified models, the efficiency of a perturbation is characterised by the count of perturbed variables (the perturbation size). However, in the case of a PSBN, a perturbation might work only for a subset of concrete BN models. To that end, we introduce and quantify perturbation robustness. This metric characterises how efficient the given perturbation is with respect to the model uncertainty. Finally, we evaluate the novel control methods using non-trivial real-world PSBN models. We inspect the method's scalability and efficiency with respect to the size of the state space and the number of unspecified components. We also compare the robustness metrics for all three perturbation types. Our experiments support the hypothesis that one-step perturbations are significantly less robust than temporary and permanent ones.
Related projects:

You are running an old browser version. We recommend updating your browser to its latest version.

More info