Artificial intelligence: Catalyzing a new era in pancreatic cancer cure

Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal human cancers, mostly due to its insidious onset that consequently leads to diagnosis at advanced stages. The complex biology of PDAC limits effective therapeutic options, yet over 90% of cases harbor KRAS mutations, making it a compelling therapeutic target. KRAS had a notorious history of being "undruggable." Emerging novel inhibitors, however, have fundamentally shifted this paradigm by demonstrating the possibility of direct KRAS targeting. Although promising in preclinical efficacy, the clinical benefit of these inhibitors was frequently curtailed by rapid and heterogeneous resistance due to the intrinsic complexity of PDAC. Concurrently, building on structure-based drug discovery, artificial intelligence (AI) is now infusing this traditional workflow with dynamic and extensive enhancements. AI-enabled approaches in molecular biology-including high-accuracy protein structure prediction, large-scale virtual screening, and the uprising of generative molecular design-have accelerated the discovery of potent and selective molecular structures. Comprehensive hybrid models and quantum-level data processing enabled the exploration of previously latent chemical spaces, thereby creating more opportunities for the search for effective PDAC therapies. Discussions of progress in KRAS inhibitors and in AI drug discovery are rising but remain largely in parallel. As a result, this review aims to bridge these two exciting fields together. Through discussions of novel strategies to target KRAS in PDAC, emerging paradigms to address KRAS resistance in clinical trials, and current innovations in AI algorithms for molecule screening and design, we hope to underscore the unique challenges and prospects of leveraging artificial intelligence to accelerate the discovery of effective therapies for PDAC.