Cellular signaling processes are driven by protein interactions, which transmit information through various mechanisms. Phosphorylation-driven activation and inhibition of proteins is a fundamental aspect of signaling, and computational models can make this knowledge actionable. However, classic models were limited by the proteome coverage of traditional biochemistry techniques and the small amount of well characterised kinase-substrate interactions.

Recent advances in phosphoproteomics have enabled measurement of tens of thousands of phosphosites per experiment at short time scales and subcellular resolution. Additionally, screenings of kinase-substrate interactions have expanded our knowledge of serine/threonine and tyrosine kinases, while computational models have improved with the application of modern transformer architectures through protein language models.

In this study, we combine these new resources with an ultra-deep characterization of the EGF signaling response using short-term phosphoproteomics. We integrate modern data and knowledge into a network model that considers the dynamic constraints measured at proteome scale. Using this model, we uncover known and propose new mechanisms associated with EGF signaling. This study not only uncovers new EGF-related signaling mechanisms, but also aims to expand the community"s understanding of the complexity of signaling pathways, demonstrating that they are at least an order of magnitude larger than traditionally described.