Optimal network control algorithms for multi-target drug design in triple-negative breast cancer

Triple-negative breast cancer (TNBC) is sustained by redundant and adaptive signaling pathways that limit the durability of single-target inhibitors. We developed a Network-Control Multi-Objective Genetic Algorithm (NC-MOGA) that integrates structural controllability, nonlinear state-transition dynamics, pathway coverage, control energy, robustness, and druggability to nominate minimal multi-target intervention sets. A curated 40-node TNBC signaling network was modeled as a constrained nonlinear dynamical system. Candidate target sets were optimized against a desired apoptotic, low-proliferation state and converted into hypothetical multi-target chemical designs. In simulated benchmarking, NC-MOGA reduced target-state error by 57.1% relative to greedy centrality and by 66.4% relative to maximum-matching control, while lowering control energy by 42.0% and 53.6%, respectively. The top candidate, NCD-01, jointly modulated EGFR, JAK2, and AXL, suppressed convergent PI3K–AKT, JAK–STAT, and YAP/TEAD signaling, and achieved median simulated IC50 values of 0.42 µM across five TNBC cell lines with a 14-fold selectivity window over MCF-10A cells. Simulated orthotopic evaluation predicted 82% tumor-volume reduction and 86% survival at day 70. These results illustrate how optimal network control can translate systems-level disease models into experimentally testable multi-target drug hypotheses.