mRNA Design Enters a More Predictive Era

New AI research forecasts mRNA protein output earlier, helping developers refine drug designs before costly lab testing begins

17 Dec 2025

Artificial intelligence is beginning to play a larger role in how mRNA drugs are designed, as developers look for ways to reduce reliance on long and costly experimental cycles.

Researchers at the University of Texas at Austin, working with French drugmaker Sanofi, have developed an AI model called RiboNN that predicts how efficiently an mRNA sequence will produce protein inside human cells. The research was published in Nature Biotechnology.

Protein expression is central to mRNA drug development. If a sequence produces too little protein, a therapy is unlikely to work. Improving performance has traditionally required repeated rounds of laboratory testing, adjusting sequences and measuring results. RiboNN aims to move part of that process upstream by forecasting performance before experiments begin.

The model was trained on large datasets from earlier mRNA studies, allowing it to learn which sequence features tend to lead to higher protein output. Researchers say it is designed to guide decision-making rather than replace laboratory work, helping scientists rule out weaker candidates earlier and focus resources on more promising designs.

According to the research team and reporting from news.utexas.edu, this approach could shorten development timelines and reduce costs by limiting the number of physical tests needed in early stages.

The development comes as mRNA technology expands beyond Covid-19 vaccines into cancer therapies and treatments for rare diseases. As these applications grow more complex, development costs and timelines have become a concern. Industry analysts say tools that support earlier and more informed design choices are gaining importance as investors scrutinise spending and scalability more closely.

Large biotechnology groups have long promoted the use of data science and digital design in drug discovery. RiboNN reflects a gradual shift in practice rather than a sudden change, and highlights the role of academic-industry partnerships in advancing such tools, even as companies pursue different strategies.

Questions remain about how broadly such models can be applied. AI systems depend on the quality and scope of the data used to train them, and regulators are still considering how AI-assisted design fits into drug approval processes.

Even so, the direction of travel is clear. As mRNA development matures, predictive tools such as RiboNN suggest that foresight, alongside experimentation, is becoming a defining feature of the field.