What problem does it solve?
Medicinal chemists and drug discovery teams need quick ADME triage on candidate molecules without running docking or providing a protein target. This Skill predicts Tier-1 ADME properties (solubility, permeability, lipophilicity/logD) directly from SMILES strings using the Boltz API, with cost estimation and confirmation before any spend.
Core Features & Use Cases
- Structure-free ADME prediction: Submit bare SMILES, CSV, .smi, or .txt molecule lists and receive per-molecule solubility (categorical), permeability (numeric), and lipophilicity (LogD) scores.
- Batch handling with cost control: Enforces the 128-molecule-per-request cap by splitting larger libraries into batches, and always runs estimate-cost ($0.01 per molecule) with explicit user confirmation before submitting.
- Per-molecule error isolation: One invalid SMILES fails only that molecule (status: failed with an error object) while the rest of the batch succeeds, and results are read from the local run.json output.
- Use Case: You have a library of 300 candidate compounds as SMILES. The Skill splits them into three batches of ≤128, quotes the $3.00 total cost, runs the predictions synchronously, and reports a merged table of solubility, permeability, and logD per compound.
Quick Start
Use the boltz-small-molecule-adme skill to estimate cost and predict ADME properties for this list of SMILES molecules.
