Advances in artificial intelligence are expanding how researchers analyze complex biological structures. Scientists at the Department of Energy’s Lawrence Berkeley National Laboratory and collaborating institutions have developed a new computational tool that combines artificial intelligence and quantum-mechanical calculations to improve protein structure determination. The research describing the approach was published in Nature Communications.
The program, known as AI-enabled Quantum Refinement (AQuaRef), helps structural biologists generate more precise molecular models by predicting the positions of atoms and electrons in protein structures. By integrating machine learning with quantum-mechanical calculations, the method produces higher-quality structural information while reducing the computational resources typically required for these analyses.
Understanding protein structures is a central task in structural biology because molecular shape and atomic interactions determine how proteins function in both healthy and diseased cells. Improved methods for protein structure determination can therefore support research in areas ranging from drug discovery to bioenergy production.
AI structural biology tool enhances protein structure determination
Traditional workflows for protein structure determination rely on two major sources of information: experimental measurements—often generated through techniques such as X-ray crystallography or cryogenic electron microscopy—and theoretical data derived from previously known protein structures.
However, these approaches can struggle to accurately model complex interactions within proteins, particularly subtle noncovalent interactions that influence molecular stability. The AQuaRef system addresses this limitation by integrating machine learning models with the Phenix software suite, a widely used platform for macromolecular structure modeling.
The combined approach enables researchers to calculate energy and force interactions within proteins more precisely, allowing structural models to better match experimental data.
“We’re all basically a bunch of proteins,” said Nigel Moriarty, a Berkeley Lab researcher involved in the study. “They do so much in our bodies that detail the processes of life. Understanding their structure can give us insights into the mechanisms that cause disease in humans or produce energy in plants. All of this knowledge can lead to more effective therapeutics and bioenergy production.”
Quantum refinement improves structural accuracy
In testing across 71 experiments, the AQuaRef approach generated higher-quality structural information while maintaining agreement with experimental measurements that was equal to or better.
The system also successfully resolved proton positions in the human protein DJ-1, which has been linked to certain forms of Parkinson’s disease and has historically been difficult to model accurately.
By enabling quantum-level refinement within practical computational limits, the tool may expand how researchers approach complex protein structures that previously required extensive manual interpretation.
Implications for structural biology laboratories
For laboratories conducting structural biology research, tools that improve protein structure determination can help accelerate model refinement and increase confidence in experimental results. AI-assisted computational methods may also support more efficient interpretation of cryo-EM and crystallography data, which increasingly generate large and complex datasets.
The researchers plan to expand the approach to additional classes of molecular structures, including those relevant to pharmaceutical research and drug design. According to the team, more detailed structural models could also support studies of plant proteins involved in photosynthesis and biofuel production.
“There is a near-infinite number of things that can benefit from a detailed understanding of these mechanisms and protein structure,” Moriarty said. “I’m excited to see how the paradigm shift that AQuaRef represents impacts the field of protein structure determination.”
This article was created with the assistance of Generative AI and has undergone editorial review before publishing.
