Proteins are the microscopic workhorses of the human body, driving everything from oxygen transport to muscle contraction. Their complex 3D structures dictate their function, and understanding these shapes is key to unlocking the secrets of life itself.
AlphaFold, an artificial intelligence (AI) tool developed by Google DeepMind, is advancing this field by accurately predicting protein structures from amino acid sequences within minutes, a task that used to take years of painstaking experimental work. The breakthrough earned AlphaFold’s creators the prestigious Nobel Prize in Chemistry in October 2024.
With more than 200 million predicted protein structures available in the AlphaFold database, it is being used by more than 2.5 million users globally – including over a million in Asia-Pacific – to advance biology and medical research. This includes a multidisciplinary research team led by Singapore’s Agency for Science, Technology and Research (A*Star) that is investigating Parkinson’s disease.
A debilitating neurodegenerative disorder affecting millions worldwide, Parkinson’s disease affects motor skills and quality of life. Early diagnosis remains a challenge, hindering timely intervention and treatment. A*Star, together with Duke-NUS Medical School and the National Neuroscience Institute, is using AlphaFold to investigate the role of a specific protein called Stip1 (stress-inducible phosphoprotein 1) in the development and progression of the disease.
Stip1, expressed in high abundance in the brain and kidneys, plays an important role in ensuring other proteins fold correctly. Elevated levels of Stip1 are found in various diseases, including Parkinson’s and Alzheimer’s. Specifically, the increase in Stip1 in Parkinson’s disease patients correlates with a rise in autoantibodies, or antibodies that mistakenly target the body’s own tissues and proteins.
During a media briefing, Jackwee Lim, a researcher at A*Star’s Singapore immunology network, explained how AlphaFold provided insights into Stip1’s structure and interactions.
“AlphaFold enabled us to create a 3D model of Stip1, showing its different domains and how they interact in a linear or non-linear manner, or in space,” Lim said, noting that model was used to map the binding sites of other proteins like HSP70 and HSP90 that work with Stip1 to facilitate protein-folding. This would help researchers understand how autoantibodies could disrupt Stip1’s normal activity, leading to misfolded protein aggregations, a hallmark of neurodegenerative diseases.
This research, still in progress, holds much potential for developing blood-based diagnostics for Parkinson’s disease. Measuring Stip1-specific autoantibodies could provide an early detection method, enabling timely interventions. The team’s work is testament to the power of AlphaFold in accelerating research timelines, providing insights that were previously unattainable.
Besides Parkinson’s disease, AlphaFold has also been deployed in a range of applications across the region. In the Philippines, for example, researchers at the International Rice Research Institute are using AlphaFold to study phosphorylation in rice, a key process that controls how rice plants respond to environmental stresses.
By predicting the 3D structures of proteins involved in this process, scientists hope to identify the molecular mechanisms that make some rice strains more resilient to drought and disease, ultimately leading to the development of stronger rice varieties that can contribute to food security.
Researchers have also used AlphaFold to understand a protein central to the immune system of honeybees, cutting research time from years to days and laying the foundation for new research to improve the survival of the keystone species that’s vital to pollination and agriculture.
Last year, Google DeepMind released a new version of AlphaFold that will enable scientists to predict the structures of more complex biological molecules, such as DNA, RNA, ligands and ions, giving them a richer and more comprehensive view of biological interactions, said Dhavi Hariharan, product manager at Google DeepMind.
The widespread availability of a powerful technology like AlphaFold, however, raises ethical questions. Hariharan said the company conducted a thorough ethical review before releasing AlphaFold, concluding that the benefits far outweighed the potential risks. It has also prioritised making AlphaFold accessible to researchers worldwide, regardless of their resources, and is actively working to provide educational resources and support to facilitate the use of the technology, she added.