Artificial intelligence is fundamentally reshaping how scientists study proteins, cutting research timelines from months or years to minutes and reducing costs by more than 60% compared to traditional laboratory methods. At a recent University of Utah symposium on AI in protein discovery, researchers demonstrated how deep learning models are enabling faster, cheaper identification of protein structures and interactions critical to understanding diseases like cancer, diabetes, and metabolic disorders.

What Is the Protein Folding Problem and Why Does It Matter?

For 50 years, biology has grappled with a fundamental challenge: a protein's function in the human body depends entirely on its complex, twisted three-dimensional shape, yet determining that shape from a simple string of amino acids has traditionally required months or even years of painstaking laboratory work. Before AI, mapping a brand-new protein structure from scratch could cost anywhere between $66,000 to $138,000, with structural biology labs estimating that over 60% of the total budget spent on traditional structure determination goes toward failed attempts.

This bottleneck has slowed medical research across the board. Understanding how proteins interact with each other and with small molecules is essential for developing new treatments, yet the sheer cost and time investment has limited how many structures scientists could realistically study.

How Are AI Models Like AlphaFold Changing Protein Research?

Deep learning models, including Google DeepMind's AlphaFold and the University of Washington's RosettaFold, have pioneered a technological leap by utilizing databases of known biological structures to train AI systems that can predict the three-dimensional shapes of proteins in a matter of minutes. Modern AI goes beyond mapping single proteins; it can now map complex interactions between multiple molecules, opening entirely new avenues for drug discovery and disease understanding.

At the University of Utah symposium, undergraduate researcher Amanda Karchner presented work demonstrating how AI can help identify protein interactions linked to diseases. Using the AI model Boltz-2 through the university's Center for High Performance Computing (CHPC), Karchner and her peers screened potential protein interactions far more efficiently than traditional laboratory methods would allow.

"When we talk about protein-metabolite interactions, those are so important for all kinds of cellular regulation. When your cellular regulation is messed up, that's cancer or metabolic disease," Karchner said during her presentation.

Amanda Karchner, Undergraduate Researcher, Department of Nutrition and Integrative Physiology, University of Utah

Rather than replacing laboratory experiments, Karchner explained that AI serves as a hypothesis-generating tool that helps researchers identify the most promising protein interactions to test experimentally. This approach dramatically reduces wasted effort on dead-end experiments.

How to Leverage AI Tools for Protein Research?

• Access High-Performance Computing Resources: Researchers can use supercomputing centers like the University of Utah's CHPC to run AI models that exceed standard laptop capacity, enabling large-scale protein structure predictions without expensive in-house infrastructure.
• Use AI as a Screening Tool First: Rather than running expensive wet-lab experiments on all potential protein interactions, use AI models to narrow down the most promising candidates, then validate only the top results experimentally.
• Combine Multiple AI Models: Different models like AlphaFold, RosettaFold, and Boltz-2 have different strengths; using multiple approaches can increase confidence in predictions and catch interactions that a single model might miss.

Martin Cuma, a computational scientist at the CHPC, explained the practical value of these resources: "We're basically helping users enable their research on our systems. Every researcher who's doing some computation on their laptop and the computation is too big for their laptop, they need to go onto a supercomputer". The CHPC provides researchers with access to supercomputers, large-scale data storage, specialized software, and technical support needed to run AI models that would otherwise be out of reach.

What Does This Cost Revolution Mean for Science?

The shift from traditional protein structure determination to AI-powered prediction represents one of the most radical cost reductions in modern scientific history. While an experimental lab requires heavy labor and thousands of dollars per attempt, with AI, if a prediction fails or has low accuracy, the cost is negligible. This economic advantage fundamentally changes the calculus of research: scientists can now afford to explore far more hypotheses and test more protein interactions than ever before.

By narrowing potential protein interactions to the most promising results, AI allows scientists to spend their time and resources testing discoveries that could genuinely improve medical understanding and lead to new treatments. The technology has made protein structure prediction reliable enough that many laboratories are now incorporating AI tools into their standard workflows, accelerating the pace of biological discovery across institutions worldwide.

As AI continues to evolve, the technology and software are helping researchers at institutions like the University of Utah accelerate discovery while reducing the time and cost required to study complex biological questions. This democratization of protein research, powered by accessible AI tools and shared computing infrastructure, is reshaping how the next generation of scientists approaches the fundamental problems of biology.