Researchers at Linköping University have made a significant breakthrough in protein structure prediction by improving the AI tool AlphaFold to predict very large and complex protein structures. This achievement, published in Nature Communications, is a crucial step towards more efficient development of new proteins for medical drugs. Proteins are essential molecules that regulate cell functions, and their shape determines their function. For over 50 years, researchers have been trying to predict and design protein structures, but it has been a laborious and expensive task. In 2020, Deepmind released AlphaFold, an artificial intelligence based on neural networks that can predict protein folding with great accuracy. However, the programme had limitations, including inability to predict very large protein compounds or draw conclusions from experimental data. Claudio Mirabello and Björn Wallner, researchers at Linköping University, have now developed AlphaFold further to overcome these shortcomings, creating a new tool called AF_unmasked that can take in information from experiments and partial data as well as predict very large and complex protein structures. This breakthrough has the potential to revolutionize protein design and development of medical drugs.
↧