A new artificial intelligence model, developed through a partnership between The University of Texas at Austin and Sanofi, aims to enhance drug and vaccine discovery by predicting the efficiency of mRNA sequences in protein production. This advancement seeks to reduce trial-and-error experimentation in developing mRNA therapeutics.
The model, named RiboNN, is designed to guide the creation of new mRNA-based treatments by indicating which sequences will produce the most protein or target specific body parts like the heart or liver. “When we started this project over six years ago, there was no obvious application,” said Can Cenik, an associate professor at UT Austin who co-led the research with Vikram Agarwal from Sanofi. “We were curious whether cells coordinate which mRNAs they produce and how efficiently they are translated into proteins.”
Funding for this work came from several sources including the National Institutes of Health and The Welch Foundation. Testing on more than 140 human and mouse cell types showed that RiboNN is about twice as accurate as previous methods in predicting translation efficiency.
The development process involved creating a dataset from over 10,000 experiments on different cell types. AI experts from both UT and Sanofi collaborated to build RiboNN. One aim is to eventually create therapies targeted at specific cell types. Cenik noted that this opens opportunities for modifying mRNA sequences to boost protein production in particular cells.
In related research published alongside their main paper in Nature Biotechnology, it was shown that mRNAs with similar biological functions are translated into proteins at comparable levels across various cell types.
Undergraduate researchers at UT played a role in ensuring data accuracy for training the AI model. Other contributors included Logan Persyn from UT and Dinghai Zheng and Jun Wang from Sanofi.
