Effective target validation hinges on directional, evidence-backed mechanistic rationales. Here we present Galactic Data V3, Biorelate's enterprise-scale causal knowledge graph platform, which converts unstructured biomedical literature into structured, audit-ready mechanistic intelligence across 50M+ documents, 3B+ biological entities, and 637K+ directional causal relationships extracted with 93% NER+linking precision.

We demonstrate the platform through a full mechanistic characterisation of Efruxifermin (EFX), a potent FGF21 agonist in clinical development for Metabolic Dysfunction-Associated Steatohepatitis (MASH). All evidence is fully traceable to retrievable primary literature evidence mapped to specific cell types and tissues, providing the mechanistic specificity required to de-risk clinical assets and rationalise combination strategies prior to experimental investment.