Using Knowledge Graphs & LLMs to Represent Scientific Knowledge

Event Description: Biology and medicine are complex topics, and it is only natural to look to breakthroughs in AI to help us navigate them. However, modern generative AI models have a complexity that rivals that of biological systems. We will discuss approaches to understanding current state-of-the-art models from the lens of traditional biomedical science and ask whether we can achieve “molecular causality” in our current age of foundation models.

Building on these advances, we will discover two open-source tools for increasing accessibility in biomedical research:

• BioCypher simplifies the organization of complex biological data into unified, accessible knowledge graphs. By streamlining data curation and fostering collaboration, it accelerates scientific discovery and makes handling vast biomedical information more manageable. (Read more about BioCypher here.)
• BioChatter takes this a step further by integrating advanced AI language models. It allows researchers to interact with these knowledge graphs using natural, conversational language, making data exploration intuitive and accessible—even for those without extensive technical expertise. (Read more about BioChatter here.)

Together, BioCypher and BioChatter empower scientists to explore complex biological phenomena, facilitate personalized medicine in areas like cancer research, and prevent reinventing many wheels via an open-source philosophy.

Join us for an exciting session of learning, exploration, and networking. Come see the potential of AI and knowledge graphs to transform data into scientific knowledge!