DeepMind, a subsidiary of Google, has developed an Alphabold programme that establishes the three-dimensional structure of proteins with a precision that has never been achieved. This is an extraordinary opportunity for medical research and drug design.
While determining the exact chemical content of a given enzyme is a rudimentary challenge these days, successfully identifying its three-dimensional shape can take years of biochemical experimentation. Several techniques exist to presumptively determine the 3D structure of a protein: X-ray crystallography, nuclear magnetic resonance, electron microscopy… all of which take several months or even years to produce a result. A London-based artificial intelligence laboratory, DeepMind, acquired in 2014 by Google, has developed a system that can automatically predict the shape of enzymes and other proteins, currently 350 000 in total, in about two hours for a chain of a few hundred amino acids. Each prediction in the database is given a confidence score indicating how accurate it is likely to be, with the system providing a good prediction 95% of the time. This represents a significant acceleration in the search. The database references the three-dimensional structures of all proteins in the human genome, as well as those found in other organisms (including mice, fruit flies and E. coli).
A useful advance in any field of biology
This huge and precise biological map could accelerate the understanding of diseases, the development of new drugs and the improvement of existing treatments. It may also be able to develop new biological tools, for example to break down plastic and recycle it into reusable materials. It may also be able to explain how bacteria are resistant to antibiotics… and how to get around this difficulty.
An open access database
To support research in biology, DeepMind researchers have decided to share their database of protein structures freely. All scientists around the world are invited to use the technology. A better understanding of proteins helps to understand and fight viruses… the possibilities are endless! The European Molecular Biology Laboratory has joined the initiative to set up this open-access database, which already contains the predicted 350,000 three-dimensional protein structures and probably several million more, soon. With this prediction of the three-dimensional structure of a protein, DeepMind will leave its mark on the history of science.
#research, #AI, #openaccess, #deepmind, #proteins,
