Empowering natural product science with AI: leveraging multimodal data and knowledge graphs.

Area of Science:

• Natural Product Research
• Cheminformatics
• Artificial Intelligence in Science

Background:

• Artificial intelligence (AI) is transforming scientific discovery across disciplines.
• The application of AI in natural product research remains limited due to data challenges.
• Existing natural product data is multimodal, unbalanced, unstandardized, and fragmented across repositories.

Purpose of the Study:

• To address the limitations of AI application in natural product research.
• To support initiatives for democratizing natural product data.
• To propose a knowledge graph approach for developing advanced AI models.

Main Methods:

• Identifying challenges in natural product data accessibility and standardization.
• Advocating for the creation of a comprehensive natural product knowledge graph.
• Exploring the potential of knowledge graphs to integrate diverse data types.

Main Results:

• The current state of natural product data hinders the development of AI models.
• A knowledge graph can unify scattered and heterogeneous natural product data.
• AI models trained on a knowledge graph could emulate human cognitive processes.

Conclusions:

• A unified knowledge graph is crucial for unlocking AI's potential in natural product science.
• Democratizing natural product data through knowledge graphs will accelerate discovery.
• Future AI models can mimic natural product scientists' decision-making with improved data infrastructure.