Scientists at Pepperdine University and UCLA are using artificial intelligence to predict kidney disease risk by mapping protein structures in minutes, a breakthrough that could accelerate drug development and reduce animal testing. By leveraging DeepMind's AlphaFold technology, researchers can now visualize how mutations in kidney proteins lead to disease, transforming a process that once required expensive microscopy into a rapid computational task.

Why Does Protein Shape Matter for Kidney Health?

Kidney proteins come in diverse shapes: microscopic corkscrews, winding ropes, and flat ribbons. Each shape determines what the protein can do. A protein might function as a channel allowing water to pass through a delicate membrane, or as a pump moving sodium. When a protein's shape mutates even slightly, the kidney loses its ability to perform that specific function, leading to disease.

The problem is that these proteins are incredibly small and easily blurred by neighboring tissues. For most of medical history, scientists could only guess at their precise structures or catch glimpses through expensive microscopic examination. Now, AI can produce a detailed 3D model in minutes, giving researchers the exact information they need to assess disease risk and design targeted treatments.

How Does AlphaFold Help Scientists Design Better Drugs?

Pepperdine alumnus Sean Wu and associate professor Fabien Scalzo recently published their findings in Nature Reviews Nephrology, exploring how AlphaFold and its successors, AlphaFold2 and AlphaFold3, treat protein folding as a pattern-recognition problem. These systems use amino acid sequences as a guiding code to predict the shape of any kidney protein, whether normal or mutated.

Drug design traditionally works like a lock-and-key mechanism: a medication must bind to a specific cavity in a protein, fitting into grooves and gaps shaped to hold a small molecule. With AI, scientists can determine the exact shape of the protein, or "the shape of the lock," making it much faster to design the right "key".

"We call this AI model the digital twin of certain organs and their protein structures. This AI modeling cuts down on trial and error. Instead of adjusting medications for side effects, in a few hours you determine that you want to target a particular gene or protein. AI will provide the structure of the compound and accordingly, the drug that you should make, which you would then test. That's the kind of pipeline we're building," said Fabien Scalzo, associate professor of computer science at Pepperdine and director of the Keck Data Science Institute.

Fabien Scalzo, Associate Professor of Computer Science at Pepperdine University

What Are the Real-World Benefits of This Approach?

The advantages extend beyond speed. AI modeling provides scientists with better insight into the problems they're targeting before designing a drug. Critically, it also reduces reliance on animal testing and years of trial-and-error with human participants, addressing ethical concerns in drug development.

• Faster Drug Discovery: AI can generate 3D protein structures in minutes instead of requiring expensive microscopy and months of research, accelerating the path from identifying a problem to designing a solution.
• Reduced Animal Testing: By validating protein structures computationally before moving to experimental phases, researchers can minimize the number of animal studies needed to develop new treatments.
• Personalized Risk Assessment: Scientists can now assess an individual's risk for kidney disease by analyzing how mutations in their specific proteins affect kidney function, enabling earlier intervention.
• Targeted Treatment Design: Understanding the exact shape of a mutated protein allows researchers to design drugs that specifically target the problem, rather than relying on broad-spectrum medications with more side effects.

What Are the Limitations Researchers Warn About?

Wu and Scalzo are careful to highlight important constraints. AI models typically generate a single static snapshot of a protein rather than capturing the full range of shapes it takes on inside a living cell. Additionally, these systems might generate overconfidence in researchers about proteins that are unusual or poorly represented in existing databases.

The authors stress that AI predictions work best as a starting hypothesis. Results must be validated against real experimental data, particularly from techniques that can capture proteins in their natural cellular environment. This means AI accelerates the research process, but it does not replace the need for rigorous experimental validation.

How to Use AI Protein Modeling Responsibly in Drug Development

• Start with AI as a Hypothesis: Use AlphaFold predictions as an informed starting point for research, not as a definitive answer. The AI output should guide experimental design, not replace it.
• Validate Against Real Data: Cross-check AI-generated protein structures with experimental techniques that capture proteins in their natural cellular environment, ensuring predictions match biological reality.
• Account for Database Limitations: Be aware that AI models perform best with proteins well-represented in training data. Unusual proteins or those from underrepresented populations may require extra scrutiny and additional validation steps.
• Combine with Traditional Methods: Use AI to reduce trial-and-error and animal testing, but maintain rigorous experimental validation pipelines to confirm that predicted structures translate to effective treatments.

"AI is not just chatbots taking over the world. In this case, scientists found a problem where AI could solve, going from sequence to structure. I think in these cases, it is extremely impactful to leverage AI to potentially save human lives," noted Sean Wu, Pepperdine alumnus.

Sean Wu, Pepperdine University Alumnus

This research builds on Pepperdine University's broader commitment to ethical AI stewardship. For the past four years, the university has hosted the Human Centered AI Conference on its Malibu campus. This year's upcoming conference will be held from September 25 to 26 alongside the inaugural Global Summit on Faith in AI, reflecting growing recognition that AI's most meaningful applications require careful ethical consideration.

As artificial general intelligence (AGI) capabilities advance, Wu emphasizes the importance of focusing on use cases like kidney disease treatment, where the path from AI model to human benefit is direct and measurable. By combining computational power with rigorous science, researchers are demonstrating how AI can address real medical challenges while respecting the limitations of current technology.