Chemprop

Research applicability

Message-passing neural networks for molecular property prediction.[4]

Getting started

Use the current v2 repository documentation (not old master-branch instructions); prepare labelled molecular data and validation splits, then train or apply a model.

Maintenance

Current repository fetched; shows v2 resources and a 2026-09-01 commit. Do not use old master instructions as current setup.

License

MIT for software; the old README distinguishes a CC0 logo. Re-check dependency and data licences; no independent license audit performed.

Workflow & human oversight

Resource-level classification is intentionally unspecified; instantiated workflows may differ. Humans define objectives, control data access, and check outputs.

Workflow level is not an evidence score; higher is not better evidence.

Approach

A Python package for training and applying message-passing neural networks to molecular property data; current v2 resources are documented in the repository.[4]

Limitations

Evaluation must account for domain shift and data leakage; training and prediction quality were not reproduced. No software execution or independent performance validation.

Reproducibility

Stored primary-source extract research-wide/candidate-evidence-067.md, accessed 2026-10-15, read for this review. No software, model or claimed outcome reproduced.

Adoption requirements

Molecular structures with property labels, Python training environment and validation splits.