Pinoy researchers exploiting Google AI system for protein structure

Proteins are essential components in all organisms. These molecules serve crucial roles in every process that occurs in living beings and understanding how they function supports various applications — from drug design and pathology to agriculture and climate change. Each protein, however, has a unique, complex 3D structure that determines its function.

Finding even one structure with traditional scientific methods is a years-long, expensive process. To accelerate it, artificial intelligence (AI) research lab Google DeepMind introduced AlphaFold.Last March 27, Google held a virtual briefing centering around this game-changing protein structure prediction AI system, with a focus on how AlphaFold 2 and its latest version AlphaFold 3 is actively being utilized in research projects across the Philippines and the Asia Pacific region.



AlphaFold’s impact AI in protein structures determinationAlphaFold 2, an AI system that predicts static protein structures down to the atomic level from its amino acid sequence, has already monumentally sped up protein structure discovery.To put AlphaFold’s impact into perspective, before AlphaFold debuted in 2020, the Protein Data Bank contained virtually all the protein structures discovered by mankind, which was roughly 180,000 structures.When Google DeepMind released AlphaFold’s open source, protein structure database in 2021, 325,000 protein structures were available for scientists to freely access.

By 2022, this database had grown exponentially to contain over 200 million structures. AlphaFold was even recognized with a Nobel Prize in Chemistry in October 2024 to acknowledge its contribution to science.AlphaFold, though, is not without limitations.

For example, AlphaFold 2 can only accurately predict static protein structures. Proteins structures are naturally dynamic. They change as proteins move throughout organisms and interact with other molecules.

It is this dynamic behavior that is the focus in fields such as drug discovery and disease research.AlphaFold 3, which launched in 2024, is similarly restricted. This update expanded AlphaFold’s functionality, enabling it to predict the static structure of more molecules including DNA, RNA, ligands, and antibodies as well as interactions between certain molecules, but it still does not accurately capture these molecules’ dynamic behavior.

That being said, AlphaFold is already helping scientists by providing low cost starting points in the early research stages. AlphaFold today has over 2.5 million users in 190 countries.

Of these, more than one third or over 1 million AlphaFold users are in Asia Pacific and 7900 users are in the Philippines.Speaking during the briefing, Google DeepMind product manager Dhavi Hariharan emphasized “AlphaFold is already allowing millions of researchers to answer pressing questions in biology, whether that’s developing plastic digesting enzymes to tackle plastic pollution, advancing progress on tackling antibiotic resistance, speeding up the development of an effective malaria vaccine, and so many more.”AlphaFold in the Philippines and Asia PacificHariharan then dove into actual cases in the Philippines and Asia Pacific.

First, she steered the briefing towards a study headed by the Philippines International Rice Research Institute (IRRI). IRRI is leveraging AlphaFold to study the structure of proteins involved in the rice phosphorylation process controls how rice functions and responds to its environment.Through this study, AlphaFold is helping IRRI scientists deepen their understanding of why some rice strains are more resilient to environmental stresses like drought and disease.

“Ultimately, the aim [of this research] is to then use this understanding to develop stronger rice varieties that can grow even in challenging conditions, contributing to food security and the UN’s zero hunger goal,” Hariharan expounded.Another Philippine study leveraging AlphaFold comes from researchers in the University of the Philippines Manila. These researchers are using AlphaFold to predict protein structures in their studies on vaccine design and deadly animal viruses.

Outside the Philippines, the spotlighted case was a multidisciplinary research aiming to advance early detection of the neurodegenerative disorder, Parkinson’s Disease, led by Singapore’s Agency for Science, Technology and Research (A*STAR) .Researchers found that an increase in the STIP1 protein in Parkinson’s Disease patients correlated to the increase in certain auto-antibodies that interfered with the normal activities of the STIP1 protein.The researchers then used AlphaFold to create a 3D structure of STIP1 so they could map how these auto-antibodies are disrupting the structure and better grasp how they affects the normal functioning of STIP1 in Parkinson’s Disease patients.

These discoveries supported by AlphaFold gave the researchers the foundation to build blood diagnostics to measure STIP1 autoantibodies in Parkinsons disease patients, which is another step in the journey towards early detection.To help more academic researchers learn how to use AlphaFold, Hariharan also added that Google DeepMind is continuously expanding its educational initiatives for this AI system through comprehensive educational guides, actively answering user questions, and holding workshops to help AlphaFold educators scale their teachings.The post Pinoy researchers exploiting Google AI system for protein structure appeared first on Newsbytes.

PH..