This web application predicts benchmark dose (BMD) values for PPARα activation by PFAS, a key molecular initiating event linked to liver toxicity. The biological response was quantified using a PPARα-dependent transactivation assay, with the benchmark response (BMR) defined as a 1.5-fold induction relative to solvent contro. BMD values were estimated through Bayesian model averaging.

To generalise beyond the tested panel of 34 PFAS, a quantitative structure–activity relationship (QSAR) model was built, where molecular descriptors derived from SMILES representations (MACCS, ECFP, Mordred, RDKit) served as inputs to predict BMD potency. The application enables screening of PFAS congeners and provides interpretable estimates of their potential to activate PPARα.

The model incorporates an applicability domain to flag predictions made for chemicals outside the structural space of the training set, supporting more reliable screening of thousands of PFAS congeners for their potential to activate PPARα.