PhaSepDB 3.0: a comprehensive knowledgebase of phase separation-related proteins from AI-assisted curation

Area of Science:

• Cell Biology
• Biochemistry
• Genomics

Background:

• Phase separation (PS) drives the formation of membraneless organelles (MLOs), essential for cellular functions.
• Understanding PS is critical for various cellular processes and pathological conditions.

Purpose of the Study:

• To present PhaSepDB 3.0, an updated knowledgebase of proteins involved in phase separation.
• To expand and enrich the database using a novel human-AI collaborative curation workflow.

Main Methods:

• Implemented a human-AI collaborative workflow integrating a large language model (LLM) agentic system with expert verification.
• Restructured the annotation framework to capture functional relevance, experimental evidence, and regulation of PS.
• Developed protein-wise summary pages synthesizing data from multiple publications.

Main Results:

• PhaSepDB 3.0 contains 3,484 expert-curated entries for 1849 PS-related proteins, doubling the previous version's content.
• The database now offers deeper insights into functional relevance, experimental evidence, and regulation of PS.
• New user-friendly web interfaces and protein-wise summary pages enhance data accessibility.

Conclusions:

• PhaSepDB 3.0 is a significantly expanded and enriched resource for studying phase separation.
• The human-AI collaborative approach enables efficient and comprehensive literature curation.
• This updated knowledgebase supports researchers in exploring the intricate basis and biological implications of PS.