Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
Molecular featurization hub (100+ featurizers) for ML. SMILES to fingerprints (ECFP, MACCS, MAP4), descriptors (RDKit 2D, Mordred), pretrained embeddings (ChemBERTa, GIN, Graphormer), pharmacophores. Scikit-learn compatible with parallelization/caching. For QSAR, virtual screening, similarity, and molecular DL.
license
Apache-2.0
Molfeat — Molecular Featurization Hub
Overview
Molfeat is a comprehensive Python library for molecular featurization that unifies 100+ pre-trained embeddings and hand-crafted featurizers under a scikit-learn compatible API. Convert SMILES strings into numerical representations (fingerprints, descriptors, deep learning embeddings) for QSAR modeling, virtual screening, similarity searching, and chemical space analysis.
When to Use
Building QSAR/QSPR models requiring molecular features as input
Virtual screening — ranking compound libraries by predicted activity
Similarity searching against molecular databases
Chemical space analysis — clustering, visualization, dimensionality reduction
Deep learning on molecules using pretrained embeddings (ChemBERTa, GIN)
Featurization pipelines integrating with scikit-learn or PyTorch
Comparing multiple molecular representations for benchmarking
For molecular manipulation and filtering use datamol instead; for substructure-based molecular operations use rdkit-cheminformatics
from molfeat.store.modelstore import ModelStore
store = ModelStore()
print(f"Total available: {len(store.available_models)}")
# Search for specific model
results = store.search(name="ChemBERTa")
for model in results:
print(f" {model.name}: {model.description}")
# View usage and load
card = store.search(name="ChemBERTa-77M-MLM")[0]
card.usage()
transformer = store.load("ChemBERTa-77M-MLM")
Common Workflows
Workflow 1: QSAR Model Building
from molfeat.calc import FPCalculator
from molfeat.trans import MoleculeTransformer
from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import cross_val_score
# Featurize molecules
transformer = MoleculeTransformer(FPCalculator("ecfp", radius=3), n_jobs=-1)
X = transformer(smiles_train)
print(f"Features shape: {X.shape}")
# Train and evaluate
model = RandomForestRegressor(n_estimators=100)
scores = cross_val_score(model, X, y_train, cv=5, scoring='r2')
print(f"R² = {scores.mean():.3f} ± {scores.std():.3f}")
# Save for deployment
transformer.to_state_yaml_file("production_featurizer.yml")
Workflow 2: Virtual Screening Pipeline
from sklearn.ensemble import RandomForestClassifier
# Step 1: Featurize known actives/inactives
transformer = MoleculeTransformer(FPCalculator("ecfp"), n_jobs=-1)
X_train = transformer(train_smiles)
# Step 2: Train classifier
clf = RandomForestClassifier(n_estimators=500, n_jobs=-1)
clf.fit(X_train, train_labels)
# Step 3: Screen library (e.g., 1M compounds)
X_screen = transformer(screening_smiles)
predictions = clf.predict_proba(X_screen)[:, 1]
# Step 4: Rank and select top hits
top_indices = predictions.argsort()[::-1][:1000]
top_hits = [screening_smiles[i] for i in top_indices]
print(f"Top 1000 hits selected from {len(screening_smiles)} compounds")
references/available_featurizers.md: Complete catalog of all 100+ featurizers organized by category — transformer models, GNNs, descriptors, fingerprints, pharmacophore, shape, scaffold, graph featurizers. Includes dimensions, dependencies, and selection guidance per category. Purely lookup-oriented content preserved as reference.
references/api_reference.md: Detailed API reference for molfeat.calc, molfeat.trans, and molfeat.store modules. Covers SerializableCalculator base class, all calculator subclasses with parameters, MoleculeTransformer methods, PretrainedMolTransformer, FeatConcat, ModelStore/ModelCard API, data type control, and PyTorch integration patterns.
Original file disposition:
SKILL.md (510 lines) → Core API modules 1-6, Key Concepts (architecture, selection guide), Quick Start, Workflows 1-3. "Choosing the Right Featurizer" → Key Concepts selection guide table. "Advanced Features" (custom preprocessing, batch processing, caching) → Recipes + Best Practices. "Common Featurizers Reference" table → Key Concepts selection guide. "Performance Tips" → Best Practices. Per-use-case disposition: QSAR Modeling → Workflow 1, Virtual Screening → Workflow 2, Similarity Search → Recipe, Chemical Space → When to Use bullet, scikit-learn Pipeline → Recipe, Featurizer Comparison → Workflow 3
references/api_reference.md (429 lines) → Migrated to new references/api_reference.md. Core patterns (FPCalculator, MoleculeTransformer, basic ModelStore) relocated to SKILL.md Core API modules 1-6. Detailed class methods, SerializableCalculator base class, PrecomputedMolTransformer, and PyTorch integration retained in reference
references/available_featurizers.md (334 lines) → Migrated to new references/available_featurizers.md. Top-level summary → Key Concepts selection guide table. Full categorized catalog retained in reference
references/examples.md (724 lines) → Fully consolidated inline: installation → Prerequisites; calculator examples → Core API 1-3; transformer examples → Core API 4; pretrained examples → Core API 5; ML integration → Workflows 1-3 + Recipes; advanced patterns (custom preprocessing, caching, chunk processing) → Recipes + Best Practices; troubleshooting → Troubleshooting table. No separate reference file needed — all content absorbed into SKILL.md sections
Retention: ~490 lines (SKILL.md) + ~170 lines (available_featurizers) + ~190 lines (api_reference) = ~850 / 1,273 original (excl. examples.md treated as consolidated) = ~67%. Including examples.md in denominator: ~850 / 1,997 = ~43%.
