lswcap
All biological processes revolve around proteins, but little has been elucidated over the past 50 years as to what a protein with a three-dimensional structure actually looks like. Understanding the three-dimensional structure of a protein will be a breakthrough in treating diseases, developing new drugs, and solving environmental problems.
There are billions of machines, or proteins, in the human body. Proteins carry blood, allow the eyes to detect light, and move the muscles. Proteins are involved in the biological process of all living things, and have a three-dimensional structure according to each function and role. However, humans have discovered more than 200 million proteins and only a few have specified this three-dimensional structure.
Solving a three-dimensional protein is like passing a thread through a bead of 20 kinds of amino acids. In order for amino acids to interfere with each other to unravel protein conformations, scientists tried to predict which proteins in amino acids had conformations. That is the protein folding problem that scientists have been undertaking for decades.
AlphaFold is a system developed by DeepMind to solve this problem. Alphafold learned from hundreds of thousands of protein conformational data identified so far. The algorithm learns from a large amount of data so that it is possible to accurately predict the three-dimensional structure of a protein from an amino acid thread.
Alpha fold prediction is expected to be helpful in various fields. Not only can disease cause prediction and drug development be accelerated, but it can also help identify plastic degrading enzymes and solve plastic contamination problems. It can also give you tips to remove carbon dioxide, a greenhouse gas, from the atmosphere.
In short, solving the protein folding problem can be of great help for scientists to better understand nature. While many scientists thought that solving the protein folding problem would be difficult in at least a few years, if not decades later, DeepMind says it has solved many protein folding problems with precision comparable to actual experiments.
The protein folding problem is conducted by CASP, the International Protein Structure Prediction Contest once every two years, and is conducting structural prediction evaluation devised by scientists. As a result of evaluating DeepMind technology as part of CASP in 2020, it surpassed other computer programs and achieved higher accuracy than conventional methods, which are cumbersome and cumbersome. The basic method of prediction was 90 points out of 100, but the average of the alpha fold score was 92.5 points.
Resolving the protein folding problem is expected to be of great help in the development of new drugs, but it cannot be seen that it will have a significant impact on the corona19 epidemic due to the late technology establishment. On the other hand, it is believed to be helpful in understanding genetic diseases such as infectious diseases and Alzheimer’s disease in the next period that may occur in the future. There is also an opinion that alphafold technology, which understands protein conformation, affects only a small part of the long process for scientists to identify diseases and develop new drugs. It is unclear how DeepMind will share its technology with researchers, and the future impact on science is not yet clear.
DeepMind has developed AlphaGo, an AI of Go so far, and is not thinking of selling products directly to other research institutes or companies. Demis Hasabis, CEO of DeepMind, said he will be releasing details on AlphaFold technology. However, the announcement is expected in 2021, and Hasabis also noted that it is seeking ways to share the technology with other scientists. Related information can be found here .
Add comment