Where AI could change the search for new treatments

Advertisement

AI could help researchers identify which genetic mutations deserve closer investigation, potentially reducing the time and resources required for early-stage drug discovery.

Nikolay Dokholyan, PhD, and his team at the University of Virginia School of Medicine in Charlottesville developed ERIS2, an AI tool designed to predict how genetic mutations affect protein stability.

The system can evaluate hundreds of mutations in less than five minutes, potentially helping researchers prioritize candidates for laboratory testing. Its potential applications range from improving the stability of therapeutic proteins to investigating mutations associated with cancer, though its predictions still require experimental validation.

In an email interview with Becker’s, Dr. Dokholyan discussed where AI could have the greatest near-term impact on drug discovery and how ERIS2 could help researchers investigate cancer-associated mutations.

Editor’s note: Responses have been lightly edited for clarity and length.

Question: What could ERIS2 change about the way researchers discover and develop new treatments, and where do you see its greatest near-term impact?

Dr. Nikolay Dokholyan: Genetic mutations, which typically alter proteins, the workers of the cell, underlie many diseases. Experimentally testing their effects on protein stability is often labor-intensive and costly.

ERIS2 offers a fast, accurate screening platform that evaluates hundreds of mutations in under five minutes. It is free to use, requires no registration and allows researchers to predict the effects of multiple mutations in a single batch.

In the near term, ERIS2 can help researchers narrow large-scale mutational scans and design more focused experiments. Researchers can use it to rank variants and prioritize those for laboratory testing, while protein engineers can shortlist mutations that may improve protein stability.

This is especially important for therapeutic proteins, such as antibodies, and industrial enzymes, for which stability is critical to successful production, storage and use.

Another application of ERIS2 is to understand protein evolution and what determines our resilience and adaptability to changing environments.

Q: Cancer is one of the diseases highlighted in the research. How could ERIS2 help scientists better understand cancer-related mutations or identify new opportunities for targeted therapies?

ND: Mutations in cancer-associated proteins can affect protein stability. ERIS2 could provide a useful platform for predicting these effects, helping scientists prioritize mutations for further experimental investigation.

This may offer insights into how cancer-associated mutations alter protein behavior and contribute to disease mechanisms, while pointing to new opportunities for targeted therapeutic research.

In a preliminary evaluation of RAF1, a cancer-associated protein involved in the RAS signaling pathway, ERIS2 performed well in ranking mutations, achieving a correlation of 0.73. These results highlight its potential for prioritizing cancer-associated mutations for further study.

Advertisement

Next Up in Technology & AI

Advertisement