Biocuration Matters: Harnessing AI to Model Biological Systems at PubMed Scale

Knowledge of how proteins functionally interact can inform studies of the dynamics of living systems – the subject of this BioMatters meeting – but much of this knowledge remains locked in peer-reviewed publications and is not accessible to researchers nor machines. Expert curation of open knowledgebases such as UniProt makes this knowledge FAIR (Findable, Accessible, Interoperable, and Reusable), but human curation is costly and difficult to scale to cover the entire body of peer-reviewed literature. In this presentation we will look at efforts to model biological systems in forms that are FAIR using resources such as UniProt, the Gene Ontology, and Rhea, how human-in-the-loop AI-assisted curation can help us build those models more efficiently, and how we can integrate biological models, knowledge, and high throughput ‘omics datasets using semantic web technologies to create new biological knowledge and insights.