Compound Insights
Most compound databases give you rows. The engine gives you narrative claims with citations — binding mode, kinase class, PK parameters, regulatory history, off-target mechanisms — synthesised from primary literature by project fred, then joined with structured records from public sources (ChEMBL, UniProt, ClinicalTrials.gov, Reactome, Monarch, Open Targets). Every fred-derived claim carries a PubMed ID or DOI.
BCR-ABL / KIT / PDGFR tyrosine-kinase inhibitor across nine approved oncology indications. The first compound run end-to-end through the engine — a data-rich worked example with off-target landscape, AOP chains, and domain-reviewed provenance.
View dossier → Second compoundOral non-peptide GLP-1 receptor agonist, FDA-approved April 2026. A deliberately data-sparse case: with thin structured records, the mechanistic inference layer carries proportionally more of the dossier — the engine's harder test.
View dossier → Third compound · investigationalInvestigational once-daily oral small-molecule GLP-1 receptor agonist, moving into Phase III for obesity and type 2 diabetes. No ChEMBL entry and a PubMed corpus of three records — the thinnest evidence base the engine has been given. Much of this dossier is an explicit no data found rather than an inference, which is the point: the interesting question is whether the engine declines where it should.
View dossier →