Pricing

Tier 1 (R² ≥ 0.60, actionable): BTK (R²=0.814), ALK (0.790), EGFR (0.780), ABL1 (0.725), CSF1R (0.675), JAK2 (0.608).

Tier 2 (R² ≥ 0.50, use with caution): FLT3 (0.560).

All predictions include confidence intervals and comparable known compounds for context.

Our best models (BTK, ALK, EGFR) achieve R² ≈ 0.78–0.81 with ~70% of predictions within a ×3 envelope. That means if the true τ_res is 60 minutes, we predict between 20 and 180 minutes 70% of the time.

This is sufficient to rank compounds and prioritize synthesis — not to replace experimental measurement. See the validation page for full scatter plots and fold-by-fold CV results.

A SMILES string — the standard text representation of molecular structure. You can copy it directly from ChemDraw, RDKit, PubChem, or ChEMBL. No 3D structure or protein structure needed.

Yes. The curated kinetics dataset (k_off, k_on, K_D measurements with provenance) is published under CC-BY-SA. You can browse it for free on this site, or download it directly. The prediction engine is the paid product — the ML models, API, and inference infrastructure.

Yes. Please cite the dataset as Pleco LLC (2026), Pleco Kineτics v1 [CC-BY-SA]. For the prediction models, cite the preprint (in preparation — contact us for a preview). Predictions include the model R² and confidence interval to include in your methods.

Yes, as part of an Enterprise engagement. We can train target-specific models on your proprietary kinetic measurements, or on additional targets not in our public set. Contact hello@pleco.dev to discuss scope and pricing.