AutoMolDesigner for Antibiotic Discovery: An AI-Based Open-Source Software for Automated Design of Small-Molecule...

Area of Science:

• Computational chemistry
• Drug discovery
• Machine learning

Background:

• Antibiotic resistance is a growing global health threat requiring novel small-molecule antibiotics.
• Existing computational tools for molecular design lack a holistic approach for antibiotic discovery.
• There is a need for efficient and accessible software tailored for small-molecule antibiotic design.

Purpose of the Study:

• To introduce AutoMolDesigner, a novel computational modeling software specifically designed for small-molecule antibiotic discovery.
• To provide a generalized framework integrating molecular generation and activity prediction for antibiotic screening.
• To offer an open-source, user-friendly tool for experimental scientists.

Main Methods:

• Development of a generalized framework with two modules: generative deep learning for molecular generation and automated machine learning for activity/property prediction.
• Utilizing pre-trained models and curated datasets for whole-cell-based antibiotic screening and design.
• Implementation of a Qt-based graphical user interface for multi-platform compatibility and ease of use.

Main Results:

• AutoMolDesigner integrates advanced computational techniques for efficient small-molecule antibiotic design.
• The software provides ready-to-use models and datasets for immediate application in antibiotic discovery.
• A user-friendly graphical interface enhances accessibility for researchers across different operating systems.

Conclusions:

• AutoMolDesigner addresses the deficit in specialized tools for small-molecule antibiotic discovery.
• The software facilitates efficient antibiotic screening and design through integrated generative and predictive modules.
• Its open-source nature and graphical interface promote wider adoption and further development in combating antibiotic resistance.