Capability details
What this helps you do
Build a molecular machine-learning workflow from featurization through data splitting, training and evaluation with DeepChem.
Catalog editorial description
View description sources
editor · 2026-09-22 — description: editorial summary of public documentation. Rolling latest tutorials, 2.8.1.dev. Not execution or scientific validation.
Research classification
Chemical sciences
View discipline evidence
editor · 2026-09-18 — 本提案聚焦分子机器学习与量子化学用途，化学为明确子用途，不否定其他研究领域。
Who it is for
- Researchers familiar with Python and supervised-learning evaluation.
How to use it
- Follow the tutorial's data-handling and feature-engineering examples.
- Choose a split, train a model and inspect held-out evaluation results.
AIPOCH has not verified execution.
Read usage documentationInputs
- Task-specific features or molecular representations, labels and a compatible model.
Outputs
- Processed datasets, fitted models, predictions and evaluation metrics.
Resource type
Tool
Access conditions
- The pinned LICENSE is MIT; training data, pretrained models and third-party frameworks need separate review.
- Dependencies include RDKit and scientific libraries; TensorFlow, PyTorch and JAX are model-dependent extras, and GPU use needs a suitable environment. No dataset or model was downloaded or validated, and generalization was not assessed.
- Community indexing of this software or analysis workflow only. External models, datasets, services and third-party components have separate terms. Installation, scientific accuracy and suitability were not validated.
- Random splitting may overestimate performance; select an evaluation strategy appropriate to the task.
- Model frameworks and datasets have separate dependencies and terms.
View usage information sources
Who it is for
editor · 2026-09-22 — audience: editorial summary of public documentation. Rolling latest tutorials, 2.8.1.dev. Not execution or scientific validation.
How to use it
editor · 2026-09-22 — getting_started: editorial summary of public documentation. Rolling latest tutorials, 2.8.1.dev. Not execution or scientific validation.
Inputs
editor · 2026-09-22 — inputs: editorial summary of public documentation. Rolling latest tutorials, 2.8.1.dev. Not execution or scientific validation.
Outputs
editor · 2026-09-22 — outputs: editorial summary of public documentation. Rolling latest tutorials, 2.8.1.dev. Not execution or scientific validation.
Access conditions
editor · 2026-09-22 — conditions: editorial summary of public documentation. Rolling latest tutorials, 2.8.1.dev. Not execution or scientific validation.
editor · 2026-09-18 — Previously reviewed catalog conditions retained, with English translation where needed; original review date preserved. Not a new license or runtime validation.
Documentation
editor · 2026-09-22 — documentation_url: editorial summary of public documentation. Rolling latest tutorials, 2.8.1.dev. Not execution or scientific validation.
Sources & versions
- Reference role
- implementation
- Last successful public check
- 2026-09-22T08:38:01.367Z
- Latest public check attempt
- 2026-09-22T08:38:01.367Z · ok
- GitHub metrics
- Stars 7,008 · checked · latest attempt 2026-09-22T08:38:01.367Z (ok)
Forks 2,314 · checked · latest attempt 2026-09-22T08:38:01.367Z (ok) - Latest default-branch commit
455d07f3e17e880a4d980b5b0365e4a02b417e14
Observed 2026-09-22T08:38:01.367Z- Source activity check
- fresh · last successful check 2026-09-22T08:38:01.367Z · latest attempt 2026-09-22T08:38:01.367Z (ok)
- Source status
- Active at last observation
- License
- MIT
- Fixed commit
455d07f3e17e880a4d980b5b0365e4a02b417e14- Reference checked
- 2026-09-18
Default-branch activity is separate from the fixed reference above. Source terms apply. A fixed commit identifies content; it does not establish scientific validity.
Related research
DeepChem is a scientific machine-learning library with utilities for datasets, feature processing, splitting and model evaluation. Molecular learning is one documented application within its broader scientific scope.
Claims & evidence
Continue the conversation
Research questions and code contributions belong with the original project. Catalog corrections can be proposed separately.