A team led by King Abdullah University of Science and Technology (KAUST) has developed an AI tool named Unify, designed to help scientists compare similar cell types across species that are evolutionarily distant. This tool could assist researchers in determining which findings from studies on animals are most applicable to human health. Their research was recently published in the journal Nature Communications. Traditionally, scientists compare cells using a technique called single-cell RNA sequencing, which identifies which genes are active in individual cells. This data can be used to create a "cell tree" that maps out different cell types and their functions within an organism. However, most existing comparison tools rely on direct one-to-one matches between genes. As species evolve, these direct matches become harder to find, which can lead to the overlooking of important biological similarities—such as those between immune cells in humans and mice, or neurons in fish and fruit flies. Unify takes a different approach by focusing on the functions that genes perform, rather than direct gene matches. Unlike conventional methods that are like dictionaries searching for word-for-word translations, Unify looks for shared biological meaning. "Unify uses AI models to analyze protein sequences and descriptions of gene functions," explains Huawen Zhong, the lead author and computational biologist at KAUST. "Genes with similar functions are grouped into units called 'macrogenes,' allowing Unify to identify cells performing similar tasks, even when their individual genes no longer match." Using this method, Unify was able to reconstruct relationships among 125 cell types from seven species that diverged over 700 million years ago. It could distinguish between identical genes, genes that evolved new roles, and genes that independently evolved to perform similar functions. In one experiment, Unify identified shared defense strategies in immune cells across multiple species—something that would be difficult to detect using traditional methods. In another, it accurately predicted how human blood cells would respond to a specific immune-signaling protein, based on the response of mouse immune cells. Across all genes, Unify's predictions were more accurate than existing methods. The KAUST team is now expanding Unify to include information about gene regulation and the positioning of cells within tissues. These additions will provide researchers with a more comprehensive view of the biological principles that are shared across life, helping to advance human health research. By combining AI with biological knowledge, Unify can identify patterns on a scale that would take researchers much longer to discover through traditional gene-by-gene comparisons. KAUST professor of marine science Manuel Aranda noted, "Much of what we know about human biology comes from studying animals like mice, but it's not always clear which findings are relevant to humans. Unify helps us determine which discoveries in model organisms are most likely to apply to humans, allowing research to be focused where it can have the greatest impact on understanding human health and disease."