DeepChem Molecular ML Skill

What it does

• Provides code patterns for the full DeepChem 2.8.0 pipeline: data loading (SMILES, SDF, protein sequences), featurization (circular fingerprints, graph convs, descriptors), splitting, training, evaluation and prediction.
• Shows how to use MoleculeNet benchmarks (Tox21, BBBP, Delaney) with their published splits.
• Covers graph neural networks (GCN, GAT, MPNN, AttentiveFP) and transfer learning with ChemBERTa, GROVER and MolFormer.
• Encodes best practices — scaffold splitting to avoid data leakage, balancing transformers for imbalanced labels, DiskDataset for memory limits — plus fixes for backend import errors and MKL conflicts.
• Ships three runnable scripts (predict_solubility.py, graph_neural_network.py, transfer_learning.py) and reference docs for the API and workflows.

Who it's for

• Cheminformatics and drug-discovery researchers predicting ADMET, toxicity or solubility.
• Data scientists entering chemical ML who need trustworthy benchmarks and baselines.
• ML engineers fine-tuning pretrained molecular models on small datasets.

Example uses

• "Train a solubility model on Delaney and predict for CCO and benzene" → runs predict_solubility.py and interprets the metrics.
• "Compare AttentiveFP against a Random Forest + fingerprint baseline on Tox21" → builds a fair scaffold-split comparison.
• "My 800-molecule activity dataset overfits badly" → recommends ChemBERTa fine-tuning or a simpler model with stronger regularization.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/K-Dense-AI/scientific-agent-skills/HEAD/skills/deepchem/SKILL.md

What I want: (describe your task here)

Install the skill I found on Claude Skill Mart.
Copy the skills/deepchem folder from the GitHub repo K-Dense-AI/scientific-agent-skills into my ~/.claude/skills/deepchem/.
When it's done, tell me in one line what this skill can do.