Abstract Motivation Genome-scale metabolic models lack explicit regulatory mechanisms, limiting their predictive accuracy for genetic interventions. Current methods for computing genetic Minimal Cut Sets either ignore regulatory networks entirely or use simplified acyclic representations that cannot capture regulatory feedback loops, ubiquitous features critical in cellular modeling. Results We developed gMISpy, a Python package that that enables efficient computation of genetic Minimal Intervention Sets (gMISs) in integrated genome-scale metabolic and regulatory networks. gMISpy incorporates cyclic regulatory logic into our previous computational framework using layered Boolean networks and BoNesis framework, resulting in a more accurate modeling of how regulatory interactions affect metabolic genes. Benchmarking across four different regulatory networks with Human-GEM showed consistent improvements in prediction accuracy, with Matthews Correlation Coefficient (MCC) gains ranging from 2.50% to 14.42%. Validation against cancer data from DepMap and Project Score confirmed that cyclic integration reduces false positives and better captures biological vulnerabilities compared to acyclic approaches. Availability https://github.com/PlanesLab/cyclic-gMISpy Supplementary information Supplementary data are available at Bioinformatics online.
Rodriguez-Flores et al. (Wed,) studied this question.