by Tryammbak Kansal
Proteins are one of the most important part of the biological system required to sustain life. It is also known as the "building block of life". Usually proteins present in our body have dozens of amino acid strings joined together to form a 3-D structure which is quite complex to understand. The study of these structures is important as structure determines the function of a protein.
Researchers generally use several experimental techniques such as x-ray crystallography or cryo-electron microscopy (cryo-EM) to know the complete structure of proteins. But these experimental processes are quite grueling work and takes a lot of time to decode the complex structures. They were only able to solve 170,000 structures out of 200 million proteins discovered in different life forms.
Researchers even tried to solve this problem using computational power but the progress was slow. But in this modern era of A. I and machine learning Google’s deep-learning program called AlphaFold was able to predict the protein structures quite accurately. Alphafold participated in a biennial protein-structure prediction challenge called CASP(Critical Assessment of Structure Prediction), where they were able to score 90 out of 100 which is as par with experimental results.
Alphafold combined deep learning with an “attention algorithm” that will first try to combine small groups of amino acids into a complex structure. They were able to predict the structure of a membrane protein from a species of archaea within half an hour onto which scientists have been working on for the past 10 years.
In the future, this tech can work with experimental techniques to quickly disintegrate complex protein structures and will help drug manufactures to quickly work upon complex structures of viruses(for e.g coronavirus) which can be fatal to us. Hence this technology will revolutionize the field of biotechnology