Connect

UVA School of Medicine scientists have developed a powerful artificial intelligence tool to determine how subtle changes in the proteins our cells make affect our health and drive disease. The tool will accelerate the development of new drugs and treatments for complex diseases, the core mission of UVA’s new Paul and Diane Manning Institute of Biotechnology.

Researchers led by Nikolay V. Dokholyan, PhD, of UVA’s Department of Neurology, have dubbed their new tool “ERIS2.” The deep-learning tool can predict how gene mutations affect the stability of cellular proteins. This stability matters because protein shape determines how well proteins can do their jobs, such as binding to enzymes or carrying out other essential cellular functions.

ERIS2 will help scientists more quickly and accurately identify which gene variants are likely to affect protein function and cause illnesses such as inherited genetic disorders or cancers. This understanding will then drive the development of new, more effective treatments.

“Understanding how mutations affect protein stability is key to understanding protein function and often disease,” Dokholyan said. “Predicting how mutations change protein stability helps us understand protein function and its relationship to disease, while also enabling us to design new proteins that can serve as biotechnological, diagnostic or therapeutic tools.”

The Power of Proteins

Scientists now use a variety of techniques to assess the effect of gene mutations on protein stability, but these techniques can be expensive and labor-intensive. They are also very challenging to apply to complex applications such as long sequences of amino acids, the building blocks of proteins. 

ERIS2, on the other hand, automates and accelerates that work. The researchers combined detailed protein measurements with other structural details and then trained their tool using a database of more than 100,000 known mutations. The availability of this vast dataset let Dokholyan and his team overcome limitations that have hampered prior computational models.

After putting ERIS2 to the test, the UVA researchers determined that it offers significantly improved accuracy for predicting mutation effects — what they label “strong predictive performance.” That will make it a powerful tool for understanding diseases and developing new treatments to benefit patients across Virginia and beyond.

Dokholyan has made ERIS2 available to the scientific community for free at https://eris2.dokhlab.org. No login or registration is required.

Finding ways to expedite lifesaving and life-changing new treatments is the goal of UVA’s new Manning Institute. The institute aims to more quickly and nimbly turn scientific discoveries into medicines and cures to improve the lives of patients around the world.

ERIS2 Unveiled

Dokholyan and his collaborators have revealed ERIS2 in a new article in the scientific journal Structure. The research team consisted of Srinivasan Ekambaram, Debarshi Kundu, Swaroop Ghosh, and Dokholyan. The scientists have no financial interest in the work.

ERIS2’s development was supported by the National Science Foundation, grants CNS-2129675 and CCF-2210963, and gifts from Intel. Dokholyan has received additional support from the National Institutes of Health, grants R35 GM134864, R01 AT012053-01 and P30 CA044579, and the National Science Foundation, grant 2040667.