'Alpha Fold' developed by DeepMind has shown a path with the power of AI to the super difficult problem of biology from 50 years ago, and research is accelerating
All biological processes revolve around proteins, but the actual shape of proteins with a three-dimensional structure has been largely clarified for the last 50 years. did not. 'Understanding the three-dimensional structure of proteins should be a breakthrough for the treatment of diseases, the development of new drugs, and the solution of environmental problems,' said many scientists, who have tackled this '
AlphaFold: a solution to a 50-year-old grand challenge in biology | DeepMind
https://deepmind.com/blog/article/alphafold-a-solution-to-a-50-year-old-grand-challenge-in-biology
AlphaFold | DeepMind
https://deepmind.com/research/case-studies/alphafold
'It will change everything': DeepMind's AI makes gigantic leap in solving protein structures
https://www.nature.com/articles/d41586-020-03348-4
You can see what the protein folding problem is and what DeepMind has solved by watching the following movie.
Protein folding explained --YouTube
Billions of 'machines,' or 'proteins,' are constantly working in the human body.
Proteins carry blood, allow the eyes to detect light, and move muscles.
Proteins are involved in all biological processes of all living things ...
Each has a three-dimensional structure according to its function and role.
However, while humans have discovered over 200 million proteins, only a handful have been identified for their conformation.
When the three-dimensional protein is loosened, it looks like a thread made of 20 kinds of amino acid 'beads'.
Since amino acids interfere with each other to make proteins three-dimensional, scientists have been trying to predict 'what kind of protein three-dimensional structure can be formed' from amino acids. This is the protein folding problem that scientists have been working on for decades.
The system developed by DeepMind to solve this problem is ' AlphaFold '
AlphaFold has learned from data on the conformation of hundreds of thousands of proteins identified so far. By learning from a large amount of data, the algorithm has made it possible to predict the three-dimensional structure of proteins from amino acid threads with high accuracy.
DeepMind predicts that AlphaFold's forecasts will be useful in many areas. Of course, it is possible to quickly predict the cause of illness and develop drugs ...
It may help identify the enzymes that break down plastic and help solve the problem of
It will also be a hint to remove carbon dioxide, which is a greenhouse gas, from the atmosphere.
In a nutshell, solving folding problems is a great help for scientists to better understand the natural world.
While many scientists believed that the folding problem would be solved 'at least years later, if not decades later,' DeepMind makes the folding problem of multiple proteins comparable to physics experiments. It states that it was solved with accuracy.
For folding problems, the 'International Protein Structure Prediction Contest' (CASP) is held every two years to evaluate the structure prediction devised by scientists. When DeepMind's technology was evaluated as part of the 2020 CASP, it not only outperformed other computer programs, but also recorded accuracy comparable to the tedious and time-consuming lab-based method. The lab-based forecast is a 90 out of 100 rating, but the median AlphaFold score is 92.5 out of 100.
'I've always dreamed of living this day,' said John Molt, a professor at the University of Maryland, who runs the CASP.
Solving the folding problem is expected to be of great help in the development of new drugs, but the late establishment of the technology is likely to have a significant impact on the pandemic of the new coronavirus infection (COVID-19). Is not ... On the other hand, some researchers find it useful in understanding the next pandemic that may occur in the future and genetic disorders such as Alzheimer's disease.
There is also an opinion that AlphaFold's technology to understand the three-dimensional structure of proteins affects only a small part of the long process for scientists to identify diseases and develop new drugs, and DeepMind developed the technology. It seems that the impact on future science is not yet clear, as it is unclear how to share it with researchers.
DeepMind is a 'laboratory' that has been developing Go AI 'AlphaGo' so far, and from that standpoint, he is not thinking about selling products directly to other laboratories and companies. DeepMind CEO and co-founder Demis Hassabis said he plans to announce details of AlphaFold's technology, which is expected to be announced in 2021. Hassabis also said he was looking for ways to share technology with other scientists.
Related Posts:
