Does it support 3D molecular features?

Yes, it includes featurizers like USR and USRCAT that can capture 3D shape similarity and pharmacophore-based features.

Can I use Molfeat with scikit-learn pipelines?

Yes, Molfeat provides MoleculeTransformer objects that are fully compatible with scikit-learn, allowing you to use them in standard ML workflows.

Which pretrained models are available through this skill?

The skill supports a wide range of pretrained models including ChemBERTa, ChemGPT, MolT5, and various Graph Neural Networks (GNNs).

Is it efficient for large datasets?

Yes, Molfeat includes built-in parallel processing capabilities via the n_jobs parameter to handle large compound libraries efficiently.

What chemical formats does the Molfeat skill support?

Molfeat primarily processes SMILES strings and RDKit molecule objects, converting them into numerical vectors for machine learning.

Molfeat Molecular Featurization

Name: Molfeat Molecular Featurization
Author: jimmc414

byjimmc414

•

324

Data Science & ML

Converts chemical structures into machine learning-ready numerical representations using over 100 specialized featurizers and pretrained embeddings.

About

Molfeat is a comprehensive hub for molecular featurization that simplifies the conversion of SMILES strings and RDKit molecules into high-dimensional vectors for scientific discovery. It provides a unified interface for traditional fingerprints, 2D/3D descriptors, and state-of-the-art pretrained embeddings like ChemBERTa or Graphormer, making it an essential tool for tasks such as QSAR modeling, virtual screening, and chemical space analysis. By integrating parallel processing and scikit-learn compatibility, it streamlines the data preparation phase for autonomous AI scientists and molecular machine learning pipelines.

Key Features

Built-in caching and error handling for robust chemical data processing.
Scikit-learn compatible transformers for seamless integration into ML pipelines.
324 GitHub stars
Unified access to 100+ molecular featurizers including ECFP, MACCS, and Mordred.
High-performance parallel processing for featurizing large-scale chemical datasets.
Support for state-of-the-art pretrained deep learning embeddings like ChemBERTa.

Use Cases

Building QSAR/QSPR models for predicting biological activity or physical properties.
Conducting chemical similarity searches and clustering for chemical space analysis.
Performing large-scale virtual screening to rank compound libraries for drug discovery.

About

Key Features

Built-in caching and error handling for robust chemical data processing.
Scikit-learn compatible transformers for seamless integration into ML pipelines.
324 GitHub stars
Unified access to 100+ molecular featurizers including ECFP, MACCS, and Mordred.
High-performance parallel processing for featurizing large-scale chemical datasets.
Support for state-of-the-art pretrained deep learning embeddings like ChemBERTa.

Use Cases

Building QSAR/QSPR models for predicting biological activity or physical properties.
Conducting chemical similarity searches and clustering for chemical space analysis.
Performing large-scale virtual screening to rank compound libraries for drug discovery.