Artificial Intelligence Successfully Predicts Protein Interactions – Could Lead to Wealth of New Drug Targets

Yeast proteins shown in different colors assemble together as two-, three-, four- and five-membered complexes like three-dimensional puzzle pieces to carry out cellular functions. An international team led by researchers at UT Southwestern and the University of Washington predicted the structures using artificial intelligence techniques. Credit: UT Southwestern Medical Center

Research led by UT Southwestern and Qian Kong

Qian Kong, Ph.D. Credit: UT Southwestern Medical Center

Dr. Kong led the study with David Baker, PhD, professor of biochemistry and postdoctoral mentor to Dr. Kong at the University of Washington prior to her appointment to UT Southwestern. Four lead authors were involved in the study, including Southwestern University of California computational biologist Jimin Pei, Ph.D.

Dr. Kong explained that proteins often work in pairs or groups known as complexes to accomplish every task required to keep an organism alive. While some of these interactions have been well studied, many of them remain a mystery. Building overall interactions—or descriptions of the full range of molecular interactions in a cell— would shed light on many fundamental aspects of biology and give researchers a new starting point for developing drugs that encourage or inhibit these interactions. Dr. Kong works in the emerging field of interaction, which combines bioinformatics and biology.

Until recently, the main hindrance to building an interaction was the uncertainty about the structures of many proteins, a problem that scientists have been trying to solve for half a century. In 2020 and 2021, a company called DeepMind and Dr. Baker’s lab released two AI technologies called AlphaFold (AF) and RoseTTAFold (RF) that use different strategies to predict protein structures based on the sequences of genes they produce.

In the current study, Dr. Kong, Dr. Baker and their colleagues expand on AI structure prediction tools by modeling several yeast protein complexes. Yeast is a common model organism for basic biological studies. To find the proteins likely to interact, the scientists first searched the genomes of related fungi for genes that had acquired mutations in a related fashion. Then they used two artificial intelligence techniques to determine if these proteins could fit together in three-dimensional structures.

Their work identified 1,505 potential protein complexes. Of these, 699 have already been structurally characterized, and the utility of their method has been verified. However, there have been only limited experimental data supporting 700 of the expected interactions, and another 106 have not been described.

To better understand these poorly characterized or unknown complexes, the University of Washington and UT Southwestern teams worked with colleagues from around the world who were already studying these or similar proteins. By combining the 3D models created by the scientists in the current study with information from collaborators, the teams were able to gain new insights into protein complexes involved in the maintenance and processing of genetic information, cellular construction and transport systems, and metabolism, (function(d, s, id){ var js, fjs = d.getElementsByTagName(s)[0]; if (d.getElementById(id)) return; js = d.createElement(s); js.id = id; js.src = "//connect.facebook.net/en_US/sdk.js#xfbml=1&version=v2.6"; fjs.parentNode.insertBefore(js, fjs); }(document, 'script', 'facebook-jssdk'));

Leave a Comment