DeepChem (Molecular ML)
暂无简介
- License 许可
- MIT
- 所属仓库
- rhCat/skillChip
- 最近更新
- 2026-07-31
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- DeepChem (Molecular ML)
- perks
- [ "predict_solubility", "train_gnn", "transfer_learning" ]
- skill
- deepchem
02 资源文件 28
- 其他blueprint.drawio2.7 KB
- 参考文档blueprint.json3.6 KB
- 资产文件blueprint.svg8.9 KB
- 参考文档index.json3.1 KB
- 参考文档ledger.json429 B
- 许可证LICENSE.txt1.0 KB
- 参考文档perks.json845 B
- 参考文档perks/predict_solubility/manifesto.json579 B
- 参考文档perks/predict_solubility/metadata.json1.3 KB
- 参考文档perks/predict_solubility/src/contracts.json573 B
- 脚本perks/predict_solubility/src/predict_solubility.py6.4 KB
- 脚本perks/predict_solubility/src/predict_solubility.sh1.5 KB
- 参考文档perks/predict_solubility/test/case.json311 B
- 资产文件perks/predict_solubility/test/fixture/molecules.csv66 B
- 参考文档perks/train_gnn/manifesto.json694 B
- 参考文档perks/train_gnn/metadata.json1.5 KB
- 参考文档perks/train_gnn/src/contracts.json760 B
- 脚本perks/train_gnn/src/graph_neural_network.py9.5 KB
- 脚本perks/train_gnn/src/train_gnn.sh1.4 KB
- 参考文档perks/train_gnn/test/case.json346 B
- 资产文件perks/train_gnn/test/fixture/molecules.csv61 B
- 参考文档perks/transfer_learning/manifesto.json714 B
- 参考文档perks/transfer_learning/metadata.json1.7 KB
- 参考文档perks/transfer_learning/src/contracts.json787 B
- 脚本perks/transfer_learning/src/transfer_learning.py13.3 KB
- 脚本perks/transfer_learning/src/transfer_learning.sh1.5 KB
- 参考文档perks/transfer_learning/test/case.json379 B
- 资产文件perks/transfer_learning/test/fixture/small_dataset.csv49 B
03 SKILL.md 正文
deepchem — DeepChem (Molecular ML)
Featurize, split, train, evaluate, and predict molecular properties with DeepChem — scikit-learn-wrapped regressors, graph neural networks, and pretrained transfer-learning models, on MoleculeNet benchmarks or custom SMILES CSVs.
What to look out for
Each tool emits one line of structured JSON (the audit + debug log) and writes its artifacts
under record_store. LOGS TO CHECK: that line + the named report + the executor run-ledger.
DeepChem loads its DL backend (torch/tensorflow/jax) and RDKit lazily; the porters degrade
gracefully (still write the named report) when those heavy libraries are absent.
Perks
| perk | tool | nature |
|---|---|---|
predict_solubility |
predict_solubility |
read-only / local — train a MultitaskRegressor (Delaney ECFP or custom CSV), evaluate, predict new SMILES |
train_gnn |
train_gnn |
read-only / local — train a graph neural network (gcn/gat/attentivefp/mpnn/dmpnn) on MoleculeNet or custom CSV |
transfer_learning |
transfer_learning |
read-only / local — fine-tune a pretrained model (chemberta/grover/molformer) on a property task |
All three perks read inputs and write a report to record_store; none mutate a remote or live
service, so each is destructive: false. They require deepchem plus a matching DL backend
(torch/tensorflow) and RDKit at run time; without those the porter still writes its report and
exits cleanly so the contract's output_exists holds.
How to use it
Pick a perk (predict_solubility, train_gnn, or transfer_learning), copy ledger.json →
task-ledger.json, fill its vars + record_store, then validate → compose → compile → oversight → executor.
Localized from K-Dense-AI/scientific-agent-skills
deepchem— MIT (see LICENSE.txt).
04 同仓库更多 Skills
Vercel Deploy
暂无简介
Audio Transcribe
暂无简介
Speech (OpenAI TTS)
暂无简介
Skill Installer
暂无简介
demo-skill
A demo skill used as a test fixture for the validate perk. Use when checking that quick_validate.py accepts a well-formed SKILL.md.
demo-skill
A demo skill used as a test fixture for generating agents/openai.yaml. Use when validating the gen_openai_yaml perk.