Artificial Intelligence-Driven Natural Product Drug Discovery: From Computational Genome Mining to Clinical...

Area of Science:

• Natural Products Chemistry
• Computational Biology
• Artificial Intelligence in Drug Discovery

Background:

• Natural products (NPs) are historically vital for therapeutics but face challenges in modern drug discovery due to complexity and low abundance.
• Advancements in multi-omics and AI offer new avenues to overcome these limitations and accelerate NP development.

Purpose of the Study:

• To present an operational, end-to-end AI-driven workflow for natural product drug discovery, bridging computational predictions with medicinal chemistry.
• To critically evaluate current AI methods and limitations while discussing emergent strategies for NP therapeutics.

Main Methods:

• Computational mining of biosynthetic gene clusters (BGCs) and metabolomes.
• Deep learning (DL) for structural elucidation and dereplication.
• Network-based target identification and generative molecular design using AI models (LLMs, diffusion models, GAs).

Main Results:

• AI integration streamlines NP discovery from BGC mining to molecular design, exemplified by AI-designed compounds like rentosertib.
• Current AI applications are largely preclinical, with limited quantitative evidence of improved clinical translation or productivity compared to classical methods.

Conclusions:

• AI offers significant potential to accelerate NP discovery, but robust validation and standardized approaches are crucial for clinical translation.
• Future directions include self-driving laboratories, foundation models for hypothesis-free exploration, and sustainability-aware AI frameworks for NP development.