About Stabilities in Boolean Networks + Applications to target genes prediction for cancer reversion and model selection
Date(s) : 07/09/2026 iCal
15h00 - 16h00
Boolean networks (BNs) are widely used to model gene regulatory networks, with attractors of the state transition graph interpreted as phenotypes or cell types; however, purely qualitative attractor analyses leave open the question of how stable these phenotypes really are. This talk presents a set of quantitative criteria — local stability, absorption potential, directional (gene-level) stability, and a global parametric score — that characterize attractor robustness under asynchronous dynamics and transient perturbations mimicking natural gene expression variability. Applied to a Boolean model of colorectal cancer tumorigenesis (Cho et al., developed with our KAIST collaborators through the ANR RepairNet partnership), these criteria identify master regulators and candidate target genes (e.g., CDX2, KLF6, YY1) consistent with experimental validation, and allow predicting the effects of gene knockouts/overexpression from wild-type perturbation data alone. Ongoing work extends this framework to model selection, using these stability scores to compare GRN models inferred from scRNA-seq data across tumorigenesis stages.
Emplacement
I2M Luminy - TPR2, Amphithéâtre Herbrand 130-134 (1er étage)
Catégories



