Integer linear programming for contrasting state interventions in Boolean networks

Drug discovery is a highly complex and time-consuming endeavor, often hindered by issues related to efficacy and safety, resulting in frequent late-stage drug attrition. Conventional strategies that rely on tightly controlling compounds' physicochemical properties have had limited success. One reason is that pathogenic cells (e.g., in heart arrhythmias or seizures) often utilize the same pathways as healthy ones. Additionally, cellular heterogeneity and circuit hijacking by cancerous cells (e.g., MAP kinase signaling) complicate selective targeting. To address these challenges, network-based approaches are gaining traction as alternatives to traditional reductionist models. In this study, we propose a scalable method rooted in integer linear programming (ILP) principles to identify minimal intervention strategies that selectively modulate common nodes in structurally similar Boolean networks. Unlike previous approaches such as minimal cut sets or elementary modes (EMs), which struggle with large networks due to computational limitations, our ILP-based method offers both efficiency and selectivity. EMs are used only post hoc to validate final solutions. We evaluate our approach across five case studies, demonstrating its ability to modulate target nodes in one network while preserving their state in others. The results suggest this framework could support therapeutic design strategies aimed at precision targeting with reduced off-target effects.