DrugTax: package for drug taxonomy identification and explainable feature extraction

Area of Science:

• Chemoinformatics
• Computational Chemistry
• Drug Discovery

Background:

• Small molecule characterization is crucial for drug discovery.
• Existing chemical taxonomies require adaptation for AI applications.
• Efficient feature extraction from molecular representations is needed for big data algorithms.

Purpose of the Study:

• To introduce DrugTax, a Python package for comprehensive small molecule characterization.
• To extend a chemical taxonomy for seamless integration with Artificial Intelligence (AI) approaches.
• To provide tools for bulk analysis, visualization, and chemical space representation.

Main Methods:

• Leveraging Simplified Molecular Input Line Entry System (SMILES) as input.
• Simultaneously extracting taxonomy information and key molecular features.
• Developing tools for bulk analysis, visualization, and similarity assessment.

Main Results:

• DrugTax provides an easy-to-use interface for small molecule analysis.
• The package enables the extraction of relevant features for AI deployment.
• It facilitates chemical space representation and molecule similarity assessment.

Conclusions:

• DrugTax is a valuable tool for chemoinformatic processing in drug discovery pipelines.
• The package simplifies the integration of chemical data into AI workflows.
• DrugTax is readily installable via PyPI and GitHub.