McGill University researchers have developed an artificial intelligence tool called DOLPHIN that can detect hidden disease markers by analyzing the intricate details of RNA within single cells. Published in Nature Communications, the study highlights how this innovation could revolutionize early disease detection and personalized treatment planning. The tool focuses on exon-level analysis of single-cell RNA sequencing data—going beyond traditional gene-level methods that often oversimplify complex biological signals. By examining how exons, the functional parts of genes, are spliced together, DOLPHIN reveals subtle changes in gene expression that are invisible to conventional techniques. “Genes are not just single units—they’re like Lego sets made of many smaller pieces,” explained first author Kailu Song, a Ph.D. student in McGill’s Quantitative Life Sciences program. “By analyzing how these pieces are assembled, we can uncover disease markers that have been missed for years.” In a key test, DOLPHIN analyzed data from pancreatic cancer patients and identified over 800 previously undetected disease markers. It successfully distinguished between aggressive, high-risk tumors and less severe forms of the disease—information crucial for guiding effective treatment decisions. Senior author Jun Ding, assistant professor in McGill’s Department of Medicine and a junior scientist at the Research Institute of the McGill University Health Centre, emphasized the tool’s potential to reduce trial-and-error in medicine. “This could help doctors match patients with the therapies most likely to work, improving outcomes and reducing unnecessary treatments.” Beyond immediate clinical applications, DOLPHIN represents a major step toward creating digital, or “virtual,” models of human cells. By generating more detailed and accurate profiles of individual cells, the tool enables researchers to simulate how cells respond to drugs and diseases in silico—before testing in labs or patients. This could significantly accelerate drug discovery and reduce costs. The team now aims to scale DOLPHIN to handle millions of cells across diverse datasets, laying the groundwork for more comprehensive and predictive virtual cell models. With continued development, the technology could become a cornerstone of precision medicine, transforming how diseases are diagnosed and treated.