Researchers have developed a new computer program that helps to decode how proteins interact, providing crucial insights for improving cancer prevention, diagnosis, and treatment methods. This innovative tool leverages artificial intelligence (AI) to create three-dimensional structures of complex protein arrangements.
An academic from the University of Missouri has engineered a computer program that unveils the complexities of protein interactions, delivering valuable information for enhancing cancer prevention, diagnosis, and treatments.
Jianlin “Jack” Cheng, a researcher at Mizzou’s College of Engineering, along with his student Nabin Giri, has introduced a tool named Cryo2Struct. This software utilizes AI to construct three-dimensional atomic representations of large protein complexes, as detailed in their recent publication in Nature Communications. The model relies on data from images captured by advanced microscopes, specifically cryo-electron microscopy (cryo-EM) images.
“Currently, Cryo-EM serves as a groundbreaking technology for unveiling the structures and arrangements of large proteins within cells,” noted Cheng, who holds the title of Curators’ Distinguished Professor of Electrical Engineering and Computer Science. “However, deriving protein structures from Cryo-EM data is a lengthy process that relies heavily on human effort, making it difficult and slow to reproduce. Our automated technology produces more accurate structures than existing approaches.”
Understanding Protein Predictions
To appreciate the importance of this research, it’s essential to grasp the fundamental role of proteins and the long-standing challenges in studying them.
Proteins are essential components of living organisms. They begin as chains of amino acids, which ultimately fold into specific three-dimensional shapes that dictate their function.
For over five decades, researchers have struggled to comprehend the mechanisms behind this folding process.
Cheng was among the pioneers in applying deep learning, a subset of AI, to tackle this issue. In 2012, he showcased an AI-driven model that successfully predicted protein structures, which laid the groundwork for transformative developments such as Google’s AlphaFold, now regarded as the most precise tool for predicting protein structures worldwide.
However, predicting a single protein structure is just one part of a larger puzzle. In reality, proteins don’t work alone; they operate in tandem as molecular machines executing complex biological tasks. Understanding how proteins interact is vital, as these interactions influence disease progression and inform scientists on the best treatment strategies.
Decoding the Mystery
Cheng’s Cryo2Struct acts like a detective solving a mystery without any clues.
The system evaluates cryo-EM images, pinpointing individual atoms and their positions within protein complexes, even in the absence of prior structural information. It then assembles these atoms into a complete 3D model of the protein complex, offering insights into protein functionality.
“Our technology allows scientists to construct a structure from cryo-EM data,” Cheng explained. “Once the structure is determined and its functions understood, it becomes possible to design drugs that can rectify any defects in the protein complex’s functionality.”
In a related study published in Chemistry Communications, Cheng and his student Alex Morehead investigated a different AI technique called the diffusion model, which examines how molecular structures develop from random noise into distinct forms. These methods can assist scientists in generating and improving small molecules, including medications, and discerning how and where these drugs interact with proteins.
“For example, if I have a drug and want to enhance its effectiveness for certain patients,” Cheng stated, “I can now employ AI to modify and optimize it.”
The interdisciplinary resources at Mizzou were crucial for facilitating this breakthrough. Cheng is affiliated with NextGen Precision Health, where he has access to Cryo-EM and advanced electron microscopy technologies.
“The collaborative opportunities at Mizzou, coupled with access to cutting-edge equipment, are exceptional,” he remarked. “At NextGen, we are united in the goal of advancing personalized health care, and innovations like Cryo2Struct will play a significant role in achieving that objective.”
