Science & Technology, Canada (Commonwealth Union) – A global team of researchers has unveiled fresh insights into the intricate three-dimensional arrangement of specific protein types using the formidable artificial intelligence tool AlphaFold2.
The impact of AlphaFold2 is likely to be significant in the years ahead. As researchers continue to explore its applications, we can anticipate accelerated advancements in fields ranging from medicine and biotechnology to environmental science and beyond. The ability to decipher protein structures with such precision has the potential to reshape our understanding of life at the molecular level.
AlphaFold2 addresses this longstanding problem by leveraging deep learning techniques and neural networks. Trained on a diverse dataset of protein structures, the AI model excels at predicting the 3D structures of proteins with remarkable accuracy. The AI model has made significant achievements by systematically predicting structures for all proteins. This comprehensive approach provides researchers with valuable insights into the commonalities and exceptions within the biological landscape.
Proteins, composed of lengthy chains of amino acids, undergo folding into three-dimensional configurations governed by precise rules. This diverse array of structures empowers proteins to execute their diverse functions. Across various organisms, spanning from bacteria to humans, proteins facilitate molecular transport, catalyze chemical reactions, serve as valves and pumps, and fulfill numerous other roles.
Researchers of the study point out that despite AlphaFold2 successfully predicting the three-dimensional structures of approximately 200 million proteins, it faced a challenge in discerning whether segments within certain proteins—referred to as intrinsically disordered regions (IDRs)—possess any structure at all, let alone being able to anticipate the form of that structure.
Alan Moses, a computational biologist and professor in the University of Toronto’s (U of T) Faculty of Arts & Science, pointed out that the question of whether IDRs possess a fixed structure or are simply ‘floppy’ components of proteins has been a prolonged discussion within the biochemistry and molecular biology community.
“We confirmed that, [while] AlphaFold2 still can’t predict the structure of IDRs very well … what it can do is tell us which IDRs are likely to have some structure – something that was previously impossible.”
Alan Moses, alongside co-authors including Reid Alderson, a post-doctoral researcher at the Medizinische Universität Graz (MUG), and Julie Forman-Kay, a senior scientist at the Hospital for Sick Children, has recently published a paper in the Proceedings of the National Academy of Sciences. The paper delves into the team’s research findings, offering insights that could enhance our comprehension of how proteins, particularly those with Intrinsically Disordered Regions (IDRs), contribute to diseases. Additionally, these findings hold promise for the development of novel drug treatments.
The team’s significance lies in being the first to systematically apply AlphaFold2, an AI model, to predict structures in IDRs across all proteins in humans and other organisms. Notably, AlphaFold2 wasn’t initially trained for such predictions, making this an unexpected and powerful application. Alan Moses draws an analogy, likening it to AI trained to drive a car attempting to navigate a bus—it may not handle the bus perfectly, but it can recognize the need for a driver. This breakthrough sheds light on the frequency of IDR occurrences, offering a crucial distinction between common occurrences and exceptions in the complex realm of biology. Overall, this innovative use of AI addresses the protein folding problem and contributes significantly to our understanding of IDRs and their involvement in various diseases.
“In the IDRs that AlphaFold2 predicts to have some structure, we’ve shown that mutations are far more likely to cause disease than mutations in other structureless IDRs,” added Moses. “This is an important advance in understanding how mutations in IDRs can cause disease, which is generally not well understood. We now believe that many of the mutations are disrupting the structure somehow.
