Google’s parent company Alphabet accommodates DeepMind, an AI lab for research bought by Google previously is functioning as an independent body. An evolutionary biologist “A game-changer” has been announced by DeepMind as a major breakthrough.
Nature was told by biologist Andrei Lupas that the perspective of medicine will change eventually. It will cause changes in bioengineering, research and everything. The protein folding problem has been solved by Alpha Fold, the AI system of DeepMind. The scientists have been vexed by this challenge for more than 50 years and have been solved.
The basic work in the cells by the machines is called proteins. The proteins get folded in three unique dimensional shapes after starting out as strings of Amino Acids. The working of the protein is determined by the 3D shape. Scientists might find it useful while developing a new drug as studying about the shape of the protein will help finding the molecule for binding. Anticipating which protein shape will be the best fit is difficult to find.
Year after year, shapes that the proteins will possibly adopt are predicted by the scientists and submitted for approval. The Critical Assessment of Structure Prediction judges the entries of the scientists who took to predicting shapes of proteins. For grown-ups, CASP is a fancy science contest. DeepMind won the contest in 2018 surprising everyone at CASP provoking melancholic researchers and making them uncomfortable. Though after their win, the problem was not solved.
With a pretty impressive accuracy and speed, Alpha Fold was able to predict which shapes the strings of amino acids will be accurate to fold up into. The mistakes that AI makes are generally by the atom’s width; otherwise, the overall performance of AI is commendable. Nature was told by co-founder and overseer of CASP, John Moult that it is a big deal that the problem could be solved in some sense. It can be beneficial for biology as the development of new drugs can be brought about. The implications that Alpha Fold provide are from future pandemics to Alzheimer’s disease. The misfolded proteins that cause Alzheimers can be cured. New treatment methods can be implemented and changed and be made in the existing drugs as well as the incorporation of a virus to improve the health quality. Lucas told New York Times that facing the future pandemics will be easier with developments in the existing drugs and medicines. DeepMind needs to share with scientists its technology and adds that ways are being assessed to do so.
By playing games, DeepMind has gained a name over the past few years. Strategy games like StarCraft and Go in which the built AI crushed pro gamers. AI is able to surpass human abilities and can be made by DeepMind, which is proved by the matches. DeepMind has elevated itself to addressing problems related to science other than playing video games. Neural networks, a type of artificial intelligence, are built by DeepMind, which the neurons of the human brain loosely inspires. Pattern detection can be done by feeding enough examples to a neural network. The shape that the strings tend to adopt is detected by AI. The human experts were not able to detect the patterns. The folding of other strings can be predicted thereafter.
In biology, the power of neural networks has been harnessed by several other labs. Data was fed on 2,335 molecules by a neural network which the AI researchers trained to have properties of antibacterial. They were successful in identifying brand new types of antibiotics by 107 million possibilities to have the properties.