ADMET overview
Screen absorption, metabolism, clearance and toxicity before committing to synthesis — hERG, DILI and mutagenicity, plus protein solubility from a sequence.
A molecule can bind perfectly and still fail as a drug — because it isn't absorbed, is cleared too fast, or is toxic. ADMET prediction screens for those liabilities before you commit to synthesis, so you filter bad actors early and optimize the survivors. Reach for it at hit triage, then again during lead optimization.
Availability
These tools are metered by credits and gated by your plan — plans set your monthly credits, available model families, and which modal-app tools you can run. See Plans & tiers.
What you can achieve here
- Predict absorption, distribution, metabolism, excretion, and toxicity
- Flag safety liabilities (hERG, DILI, mutagenicity) early
- Predict aqueous solubility
- Profile off-target / kinome selectivity
For which tool fits each, see What you can achieve.
Tools in this section
| Tool | What it does |
|---|---|
| ADMET-AI | Predict 41 TDC ADMET endpoints and 8 RDKit descriptors from SMILES with Chemprop-RDKit graph neural networks — hERG, AMES, DILI, CYP450 and clearance. |
| ADMETlab 3.0 | Deprecated. The external ADMETlab 3.0 service is unavailable; use ADMET-AI instead. |
| DeepChem | Retired. Use ADMET-AI for property and ADMET prediction. |
| DeepPurpose | Train and run DeepPurpose models for drug-target prediction on the platform. |
| DeepSol on the platform | Predict protein solubility for E. coli expression from an amino acid sequence. |
| GraphSol | Predict protein solubility from sequence with an attentive graph convolutional network over predicted contact maps — a probability plus a binary call. |
Recipes using these tools
Screen ADMET
Screen ADMET: a screening recipe chaining admet_ai, deeppurpose.
Head-to-head safety comparison
Compare two or three drugs on off-targets, real-world adverse-event signals, and pharmacogenomics in one pass.
Read next
Reading a result against the bench
What each ADMET score was graded against, and which endpoints translate.
What you can achieve
The outcomes these tools deliver — and which to pick when several do the same thing.
Which tool?
Match a scientific goal to a tool across every domain.
Concept clusters
How tools, jobs, and results fit together.
Workflows
Guided pipelines and copy-and-adapt prompt recipes for common goals.
Was this page helpful?