Researchers at King Abdullah University of Science and Technology have built a new AI tool called Unify. It helps scientists compare cells across different species.
The team published its findings in Nature Communications on September 21, 2026. The study focuses on how cells evolve over hundreds of millions of years.
Unify connects 125 cell types from seven species. These species span more than 700 million years of evolution.
The tool helps researchers figure out which discoveries from animal studies might apply to human health. This is a common challenge in biology research.
Scientists often study mice, fish, flies, and worms. These animals share biological functions with humans, even when their genes look different.
How Traditional Tools Fall Short
Most cell comparison tools rely on single-cell RNA sequencing. This method shows which genes are active in a cell.
Researchers use this data to build a cell tree. The tree maps different cell types and how they behave.
Older tools search for direct one-to-one gene matches between species. As species evolve apart, these matches get harder to find.
This means scientists can miss real similarities. Immune cells in humans and mice, or neurons in fish and flies, might work the same way even if their genes don't match.
How Unify Compares Cells Differently
Unify takes a different approach. Instead of matching genes directly, it looks at what job each gene performs.
Lead author Huawen Zhong, a computational biologist at KAUST, compared the method to translation. Old tools work like a dictionary that needs word for word matches. Unify looks for shared meaning instead.
The tool uses AI models trained on protein sequences and gene function descriptions. It groups genes with similar jobs into units called macrogenes.
This grouping lets Unify spot cells that do similar work, even when their genes evolved separately. The team tested this by studying immune cells across several species.
Unify found shared defense tactics that older tools missed. It also predicted how human blood cells would respond to an immune signaling protein, based on data from mouse immune cells.
The tool's predictions were more accurate than existing methods across the genes tested.
The KAUST team plans to expand Unify further. They want to add information about gene regulation and where cells sit within tissues.
KAUST Professor Manuel Aranda said much of what scientists know about human biology comes from animal studies. He said Unify helps identify which of those findings are most likely to apply to humans.
The goal is to help researchers focus their work where it matters most for understanding human health and disease.
