Using barcodes and AI to understand and control blood aging

Researchers from the Lars Velten lab at the Centre for Genomic Regulation (CRG) have achieved two major breakthroughs that bring us closer to understanding and potentially reversing the effects of aging in human blood. In a pair of studies, the team has combined cutting-edge insights from aging biology and artificial intelligence to map how our blood changes over time and how we might one day reprogram it.

In a study published in Nature, the CRG team – in collaboration with Alejo Rodriguez-Fraticelli’s group at the IRB Barcelona – unveiled how the diversity of blood stem cells declines dramatically with age. They traced natural “barcodes” in DNA (epigenetic marks like methylation that are passed from a stem cell to its daughters) to reconstruct the blood stem cell family tree in both mice and humans. For that, they developed a new technique, EPI-Clone, which reads methylation barcodes from individual cells. They found that by age 50, most people begin losing stem cell diversity, with a few dominant clones taking over – especially those biased toward producing myeloid cells linked to inflammation. This shift, which becomes nearly universal after age 60, is tied to chronic diseases like leukemia, heart conditions, and reduced immune resilience.

By the age of 60, most people have lost blood stem cell diversity in favour of a few dominant clones linked to inflammation.

Meanwhile, in an article in Cell, the Velten lab explains how they took a proactive leap: using generative AI to design synthetic DNA sequences that can control gene expression with unprecedented precision. To develop their AI model, the researchers first spend over 5 years carrying out thousands of experiments with lab models of blood formation so they could decipher the “grammar rules” of enhancers – key regulators of gene activity. They generated more than 64,000 synthetic enhancers, each designed to test binding sites for 38 different transcription factors. Their model then used the understanding of this “language” to create entirely new sentences – enhancers not found in nature, generated from scratch to activate or silence genes in specific cell types. Indeed, in experiments in mouse blood cells, these synthetic enhancers performed exactly as predicted. “It’s like writing software but for biology, a new way of giving instructions to a cell with unprecedented accuracy,” says Robert Frömel, first author of the study.

The team has managed, for the first time, to generate AI-designed synthetic molecules that can successfully control gene expression in healthy mammalian cells.

Together, these two studies mark a major advance in “generative biology.” By revealing how the blood system degrades with age and offering a way to rewrite its genetic control circuits, CRG scientists are building a foundation for precision anti-aging medicine.