Function SIG Meetings:

Bio: Dr. Cathy Wu is the Unidel Edward G. Jefferson Chair in Engineering and Computer Science and Director of the Data Science Institute at the University of Delaware. She has conducted bioinformatics data science research for 35 years and has been elected a Fellow of ISCB, ACM, IEEE, and AAAS. She has published more than 320 scientific publications with over 92,000 citations and an h-index of 83 based on Google Scholar. As the Director of the Protein Information Resource, she co-founded the UniProt Consortium in 2002 with continuous NIH funding as an MPI to provide a central international hub of protein sequence, function, and knowledge.

Abstract: We have witnessed the transformative impact of artificial intelligence and accelerated biomedical knowledge discovery with advancements in large language models (LLMs), retrieval-augmented generation (RAG), knowledge graphs (KGs), and agentic AI. In this talk, I will highlight several areas of AI development in the UniProt Consortium for protein functional annotation and biomedical discovery. These include: (1) using protein language models (pLMs) as a scalable approach for protein name and function prediction, (2) combining text mining with LLMs as a generalizable framework for both relation extraction and summarization from scientific literature, and (3) developing a protein knowledge network (ProKN) in partnership with the NIH Common Fund Data Ecosystem as an interoperable and sustainable means for data integration and knowledge discovery. All functional annotations generated by the pLMs, text mining tools, and LLMs are clearly labelled as AI annotations with evidence tagging and literature reference. The ProKN knowledge graph integration expands the utility of interconnected biomedical data for drug repurposing, understanding disease mechanisms and functional genomics. Collectively, the AI-driven approach allows the UniProt Consortium to improve the functioning of the UniProt knowledgebase while making UniProt data AI-ready to democratize access by the broad user community.