What are the common fingerprints available in Molfeat?

It includes standard fingerprints like ECFP (Circular), MACCS keys, MAP4, and pharmacophore-based featurizers like FCFP and CATS.

Does this skill support deep learning models?

Absolutely. It includes pre-trained transformers for state-of-the-art models like ChemBERTa, ChemGPT, and various Graph Neural Networks (GNNs).

What molecular formats does Molfeat support?

Molfeat primarily works with SMILES strings and RDKit molecule objects, converting them into numerical vectors or embeddings for machine learning.

How does Molfeat handle large datasets?

It features built-in parallel processing and caching mechanisms to handle high-volume molecular featurization efficiently across multiple CPU cores.

Can I use Molfeat with scikit-learn?

Yes, Molfeat provides MoleculeTransformer classes that are fully compatible with scikit-learn's Pipeline and Transformer interfaces for easy model building.

Molfeat Molecular Featurization

Name: Molfeat Molecular Featurization
Author: Zehong-Wang

byZehong-Wang

0•

Data Science & ML

Converts chemical structures into high-quality numerical features for molecular machine learning and cheminformatics tasks.

Molfeat is a comprehensive molecular featurization hub that unifies over 100 pre-trained embeddings and hand-crafted featurizers into a single, scikit-learn compatible interface. It allows researchers and developers to easily transform SMILES strings or RDKit molecules into numerical representations for QSAR modeling, virtual screening, and deep learning. By providing access to state-of-the-art models like ChemBERTa alongside traditional fingerprints like ECFP, Molfeat streamlines the data preparation phase of chemical AI projects, offering built-in parallelization and caching for high-performance chemical space analysis.

Key Features

01Access to 100+ molecular featurizers including ECFP, MACCS, and RDKit descriptors.

02Scikit-learn compatible transformers for easy inclusion in machine learning pipelines.

03Seamless integration with pre-trained deep learning models like ChemBERTa and Graphormer.

04Flexible handling of molecular data with automated error management and standardization.

050 GitHub stars

06Built-in support for parallel processing and efficient caching of expensive embeddings.

Use Cases

01Performing large-scale virtual screening and structural similarity searches.

02Building QSAR/QSPR models for property prediction and drug discovery.

03Generating embeddings for deep learning architectures trained on chemical data.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add zehong-wang/kosmos molfeat

For use in Claude.ai and ChatGPT

Download Skill