Andrew Ho, a researcher at the Institute of Myology, is the first co-author of an article that has just been published in the June 2025 issue of Cell Stem Cell*. The study reveals that a single injection of PGE2, a lipid naturally produced by the body in response to muscle injury or exercise, not only improves muscle strength, but also reactivates ‘youthful’ epigenomic programmes by reversing transcriptional dysregulation in aged muscle stem cells. While this biological finding is remarkable, the method developed to demonstrate it, which combines cutting-edge tools and artificial intelligence (AI) algorithm, is the most significant advance in this publication. The team is pioneering the use of AI in the field of muscle research.
Decoding complex biological processes with the help of AI
Through their innovative approaches, Andrew Ho and his colleagues’ work decodes complex biological processes and predict functional targets that could improve muscle function. The researchers used AI-based models such as ChromBPNet and Dual-FLIT, in combination with single-cell multi-omic analyses, to predict and validate the impact of ageing on muscle regeneration circuits and the ability of PGE2 treatment to reset them.
They started by profiling muscle stem cells in young, aged and treated subjects, then leveraged custom computational scripts to train the AI model by teaching it the “grammar” of the DNA regulatory signals that govern repair. In other words, the system learned which combinations of signals permitted or prevented gene expression, and how these patterns changed in diseased muscle or in muscle restored by therapy.
Tracking cell behaviour in real time using bioengineering tools
Moreover, the researchers combined genetic analysis with bioengineering tools: single-cell cultures in microwells functionalized with customised surface protein factors that enable robust growth and reat-time imaging of cellular fates. This approach allows the team to directly observe cell behaviour. Specifically, this system enabled the researchers to capture the growth dynamics of cultured cells in vivo and measure the effectiveness of a treatment. In practice, this makes it possible to determine more quickly whether a therapy actually corrects a defect in a micro-culture model, even before final testing; thus the researcher can make earlier and better-informed decisions, flag dosage or timing errors, and predict whether the culture is on the correct path well before in comparing with traditional testing.
This finding, which links the mechanism of molecular reset to functional recovery, provides proof of concept for the approach driving the AVATAR platform, led by Andrew HO at the Institute of Myology. Established in September 2024, AVATAR integrates advanced bioengineering tools and microfluidic technology to model muscle disease using patient-derived cells. By enabling real-time analysis of co-cultures with multiple cell types that mimic the dynamic environment of muscle tissue in vivo, the platform builds directly on this study’s foundations. Its broader strategy is to combine bioengineering with AI-driven analysis to unravel disease mechanisms, guide treatment evaluation, and establish a scalable framework that opens new avenues for precision therapies across a wide spectrum of muscle disorders.
