Corollary

Research

  • Papers

Library

  • Catalog

Account

  • Jobs
  • Billing
SettingsResearch demo · not for clinical or commercial use
Corollary
  1. Catalog
  2. Chemistry & ADMET
  3. Chemprop D-MPNN — toxicity panel

Chemistry & ADMET · Swanson et al., Stanford · Chemprop, MIT

Chemprop D-MPNN — toxicity panel

The toxicity half of the ADMET panel: Ames, hERG, DILI, ClinTox, LD50 and all twelve Tox21 assays.

Commercial use permittedThe MIT licence allows use in commercial research and products.MIT
Availability
On demand
Typical latency
900ms
Credits per run
~6
Compute tier
A · Fast

1–30s

Runs on On-demand GPU

Runs on our servers with on-demand compute. A first run needs time to load the model; active capacity can be reused and scales down when idle.

Run this model

Chemprop D-MPNN — toxicity panel

Canonical SMILES for the small molecule.

Needs ligand smiles

About this model

Directed message-passing networks are the dependable default for molecular property prediction, and this is the toxicity read of the same ADMET-AI ensemble: mutagenicity, cardiotoxicity, liver injury, clinical-trial failure, acute toxicity, the twelve Tox21 nuclear-receptor and stress-response assays, and PAINS, Brenk and NIH structural alerts. Use it to triage a series for liabilities; use ADMET-AI when you want the absorption and metabolism picture with it.

Standardized I/O contract

Every model in the Hub speaks the same contract, which is what lets the Router and the agent call any of them without special-casing.

Inputs

  • smilessmiles · required

    Ligand SMILES

Outputs

  • predictionstable

    Toxicity endpoints with percentile against approved drugs.

Specification

Hardware
1× A100 16GB
GPU memory
CPU only
Version
2.0.1
Licence
MIT
Backend
On demandOn-demand GPU
MCP server
mcp-molecule-server

Tasks

toxicityproperty prediction
chemistrytoxicitygnnmit-license