PanOnto and PanKG: A Stage-Aware Ontology and Knowledge Graph for Pancreatic Disease Progression

PanOnto represents, to our knowledge, the first formal ontology that explicitly models the continuum of pancreatic disease progression, spanning acute pancreatitis (AP) to chronic pancreatitis (CP), precursor neoplasms (PanIN/IPMN), and pancreatic ductal adenocarcinoma (PDAC). Building on this schema, PanKG is a multi-source biomedical knowledge graph that integrates ten major databases—including PrimeKG, Comparative Toxicogenomics Database, Hetionet, Reactome, DisGeNET, DrugBank, STRING, Genomics of Drug Sensitivity in Cancer, ChEMBL, and GWAS Catalog—into a unified structure comprising 334,934 nodes and 12,946,163 edges. The graph is organized across 14 Biolink-compliant entity types and 31 relationship types, and incorporates disease-stage annotations for 101 pancreatic disease–associated genes, enabling stage-aware mechanistic inference across the pancreatic disease continuum.