ParetoMol: a free web application for multi-objective pareto analysis of molecular safety and pharmacokinetics
Ilkham Yabbarov, Rodrigo A. Vargas-Hernández · Digital Discovery · 2026
Molecular optimization requires balancing conflicting objectives—potency, selectivity, pharmacokinetics, and safety—yet freely available cheminformatics tools present these properties in isolation, forcing researchers to reconcile trade-o!s manually. Here, we present ParetoMol, an open-source web application that computes Pareto dominance across user-defined molecular objectives and provides 14 synchronized visualization tabs for interactive exploration of compound trade-o!s. ParetoMol accepts molecules via SMILES strings, CSV/SDF file upload, ChEMBL target identifiers, or PubChem name resolution; additional experimental data can be merged at any point via CSV/TSV upload. Visualization tabs include configurable Pareto scatter plots, BOILED-Egg partitioning, sca!old analysis, chemical space projection (UMAP/PCA/t-SNE), activity cli! detection, parallel coordinates, radar charts, and multi-parameter optimization scoring. Three demonstrations illustrate the platform’s analytical capabilities: (i) safety-augmented Pareto analysis of 1,949 FDA-approved drugs expands the Pareto front 3.6-fold, rescuing 131 compounds with favorable predicted safety profiles; (ii) sca!old-level analysis of 50 EGFR inhibitors reveals universal predicted AhR activation across the 4-anilinoquinazoline class (100% above threshold vs. 50% in matched controls; p = 5.61x10 -10 ); and (iii) cross-class comparison of 92 compounds across six drug classes reveals class-specific predicted risk profiles with orthogonal liability axes—including NSAIDs with minimal predicted cardiac and mutagenic risk but second-highest predicted DILI—that are invisible to single-endpoint screening. ParetoMol is freely available at https://paretomol.com under the MIT license.