AI and the World's Hard Problems · Medicine & drug discovery · Nature (computational biology) · 2026-10-22
LLM-powered construction of antibiotic resistance datasets and knowledge graphs for accelerated antibiotic resistance gene discovery
To address outdated annotations, limited coverage, and high false-positive rates in antibiotic resistance research, the researchers developed an LLM pipeline that extracts resistance knowledge from full papers and supplementary materials. They built AResDB from over 8,570 papers and AResKG with over 52.1 million triples. Compared with existing knowledge graphs, AResKG improved the performance of commonly used machine learning methods for predicting gene–antibiotic resistance relationships.