Alphabet-owned DeepMind has claimed that research powered by artificial intelligence (AI) can help to predict the 3D structure of a protein on just its genetic sequence.
DeepMind further claimed that its AI-driven system, AlphaFold, is capable of generating 3D models of proteins that are highly accurate and potentially capable of solving one of the key challenges in biology.
The company is hoping that artificial intelligence research can drive and fast track new scientific discoveries.
The UK-based AI company said that for several years, scientists have found it difficult to understand the 3D shape of a protein from just its genetic sequence.
The problem is that DNA contains only information about the sequence of amino acid residues which are a protein’s building blocks that form long chains.
Predicting how the chains fold into the intricate 3D structure of a protein is an uphill task, especially when the size is bigger. It is also not easy to model big proteins as there are more interactions between amino acids that need to be noted, said the company.
According to DeepMind, prediction of the shape of a protein is vital to understanding its role within the body and also for diagnosis and treatment of diseases like Alzheimer’s, Parkinson’s, Huntington’s and cystic fibrosis that are believed to result from misfolded proteins.
DeepMind, in its blog, wrote: “We are especially excited about how it might improve our understanding of the body and how it works, enabling scientists to design new, effective cures for diseases more efficiently.
“As we acquire more knowledge about the shapes of proteins and how they operate through simulations and models, it opens up new potential within drug discovery while also reducing the costs associated with experimentation.”
DeepMind said that its team particularly worked on the complex problem of modeling target shapes from scratch, without the use of previously solved proteins as templates.
The AI company claimed that it could come up with a high degree of accuracy when predicting the physical properties of a protein structure, after which it used two distinct methods to build predictions of full protein structures.