A new artificial intelligence model developed by researchers at The University of Texas at Austin aims to improve treatments and preventive strategies in medicine. This AI model, named EvoRank, leverages evolutionary processes to design protein-based therapies and vaccines.
EvoRank represents a significant step forward in how AI can influence biomedical research and biotechnology. Researchers presented their findings at the International Conference on Machine Learning and published a related paper in Nature Communications.
A key challenge in designing better protein-based biotechnologies is the lack of sufficient experimental data about proteins. EvoRank addresses this by using natural variations of millions of proteins generated over time to extract useful dynamics for biotech solutions.
“Nature has been evolving proteins for 3 billion years, mutating or swapping out amino acids and keeping those that benefit living things,” said Daniel Diaz, a research scientist in computer science and co-lead of the Deep Proteins group at UT. “EvoRank learns how to rank the evolution that we observe around us, to essentially distill the principles that determine protein evolution and to use those principles so they can guide the development of new protein-based applications, including for drug development and vaccines, as well as a wide range of biomanufacturing purposes.”
UT houses one of the leading programs for AI research in the country and hosts the National Science Foundation-funded Institute for Foundations of Machine Learning (IFML), led by computer science professor Adam Klivans. Recently, the Advanced Research Projects Agency for Health announced a grant award involving Deep Proteins and vaccine-maker Jason McLellan from UT's molecular biosciences department. The team will receive nearly $2.5 million to apply AI in developing vaccines against herpesviruses.
“Engineering proteins with capabilities that natural proteins do not have is a recurring grand challenge in the life sciences,” Klivans said. “It also happens to be the type of task that generative AI models are made for, as they can synthesize large databases of known biochemistry and then generate new designs.”
Unlike Google DeepMind’s AlphaFold which predicts protein shapes based on amino acid sequences, EvoRank suggests alterations in proteins for specific functions such as improving their suitability for biotechnologies.
McLellan’s lab is already testing different versions of viral proteins based on AI-generated designs.
“The models have come up with substitutions we never would have thought of,” McLellan said. “They work, but they aren’t things we would have predicted, so they’re actually finding some new space for stabilizing.”
Protein therapeutics often present fewer side effects compared to alternatives and are expected to grow significantly over the next decade. However, developing these drugs remains slow, costly, and risky with high chances of failure during clinical trials.
If commercialized successfully along with its related framework Stability Oracle created by UT researchers, EvoRank could reduce time and costs associated with drug development while providing better designs faster.
Using existing databases of naturally occurring protein sequences from various organisms like starfish or oak trees allows researchers to train EvoRank effectively. By comparing different versions of proteins across species, they identify useful amino acids selected by evolution over time.
Diaz plans further developments including a multicolumn version capable of evaluating multiple mutations simultaneously affecting protein structure stability along with tools predicting relationships between structure function within these molecules.
Besides Klivans Diaz contributors include computer science graduate student Chengyue Gong UT alumnus James M Loy Tianlong Chen Qiang Liu Jeffrey Ouyang-Zhang David Yang Andrew D Ellington Alex G Dimakis Funding came from NSF Defense Threat Reduction Agency Welch Foundation
###