AI Tools and Partnerships Advance Precision Medicine
Yale researchers built genomic AI models for personalized medicine. The University of Liverpool and BPGbio will expand AI drug discovery via the NAi platform.
Researchers at Yale University have developed new artificial intelligence tools for personalized medicine, while the University of Liverpool has launched a collaboration with Boston-based biotech company BPGbio to accelerate AI-driven drug discovery and precision healthcare.
At Yale, two researchers from a university laboratory used the NCSA Delta supercomputer to build tools that account for individual genetic variations. One tool, a genomic language model called UKBioBERT, was pre-trained on real genetic variants from approximately 300,000 individuals in the UK Biobank. By combining UKBioBERT with state-of-the-art architectures, the researchers created UKBioFormer and UKBioZoi to improve gene expression predictions across individuals or genes. The models are designed to help doctors understand how individual genes might affect disease risk or drug response, particularly for conditions like cancer, diabetes, Alzheimer's and autoimmune disorders, which are driven by subtle changes in gene expression. The work was published in npj Artificial Intelligence. Using NCSA resources, the researchers reduced training time for a model with 300 genes from more than one month to 10 days.
A third-year Ph.D. student in the same lab has developed another tool called STORM (Spatial Temporal multi-Omics Representation Model), which uses graph neural networks to integrate spatial location, timing, and multiple types of genetic data into a single view. The tool aims to illuminate developmental trajectories in tissues and could enable targeted screenings or early interventions, with potential application to human atlas data.
Separately, the University of Liverpool and BPGbio announced a collaboration to accelerate drug discovery using AI. The partnership, facilitated through the University of Liverpool Civic HealthTech Innovation Zone (CHI-Zone) programme, emerged from a meeting during a Liverpool City Region trade mission to Boston in October 2024. The project will use advanced AI techniques to analyze large-scale healthcare data, including medical scans, lab results, and biological information about genes and proteins, to uncover cause-and-effect relationships in disease processes.
BPGbio will expand its NAi Bayesian AI platform through the collaboration. The platform is designed to identify underlying drivers of diseases, and the partnership aims to enhance its speed, scalability, and application in drug discovery and precision healthcare. According to BPGbio's president and CEO, NAi has already helped discover actionable drug targets and advance new therapies in oncology, mitochondrial disease, and neurology. The project will involve a multidisciplinary team of researchers, including experts in public health data science, Bayesian inference, proteomics, and AI translation.
The collaboration is expected to create jobs in the region and strengthen Liverpool's position in the global health-tech sector. CHI-Zone is funded by the Liverpool City Region Life Sciences Innovation Zone Programme, part of the Government's national Investment Zone Programme.