[ad_1]
Image source,CASP / DEEPMIND / VTAGLIABRACCIDTOMCHICK, SOUTHWEST
DeepMind Predicts Protein Model of Legionnaires’ Disease Bacteria
For half a century, predicting how a protein folds into its unique three-dimensional shape has puzzled scientists and is one of the main challenges in biology.
But experts announced that one of the biggest mysteries in the biological world has now basically been solved by artificial intelligence AI.
DeepMind (Deep Thinking), a British artificial intelligence company based in London, said they “basically solved this problem.”
Better understanding and prediction of protein shapes will play a key role in the development of new drugs in the future. Technological advancements made by DeepMind, a Google subsidiary, are expected to accelerate research on a variety of diseases, including the new crown.
Independent scientists in the United States stated that the precision of the DeepMind system in predicting protein shape is comparable to that of expensive and time-consuming laboratory methods.
Dr. Andriy Kryshtafovych of the University of California, Davis, who is a member of the scientific jury, described this DeepMind achievement as “excellent.”
Dr. Krestavovich said that quickly and accurately finding out the shape of proteins has the potential to revolutionize the life sciences.
What is the shape of protein?
Image source,fake images
The computer displays the protein that binds to the receptor
Protein exists in all organisms and is the basis for cell survival. They play a central role in the chemical processes essential for life.
Proteins are composed of polypeptide chains formed by combining amino acids in a certain order, which are folded into various delicate shapes in countless ways, and this is the key to their important role in the body.
Many diseases are related to the role that proteins play, for example, they can become enzymes that trigger chemical reactions, antibodies that fight diseases or the hormone insulin as a chemical messenger.
Dr. John Moult of the University of Maryland is chairman of the scientific jury. He explained that even the smallest rearrangement of these protein molecules would have a catastrophic impact on people’s health. Therefore, to understand diseases and find new treatments, it is necessary to study proteins.
Mort further explained that there are thousands of human proteins and billions of proteins in other species, including those of bacteria and viruses. Today, just trying to figure out the shape of a protein takes years of time and expensive equipment.
Prediction contest
In 1972, the American biologist Christian Anfinsen won the Nobel Prize for his research on the relationship between the amino acid sequence and the biologically active conformation.
Anfinson believes that it should be possible to determine the shape of a protein based on the sequence of its constituent amino acids.
Since then, every two years, dozens of research teams in more than 20 countries around the world have tried to get computers to predict the shapes of about 100 proteins through amino acid sequences.
At the same time, biologists use traditional techniques such as X-ray crystallography and nuclear magnetic resonance spectroscopy (NMR spectroscopy) to determine the three-dimensional structure of proteins in the laboratory to understand each protein molecule. The relative position of an atom.
After that, a dedicated CASP team made up of scientists (equivalent to the Community Experiment on Critical Evaluation of Techniques for the Prediction of Protein Structure) used the computer to predict the three-dimensional structure of the protein and use the laboratory. The results of the three-dimensional structure obtained by the method are compared.
The CASP review team used a 0-100 measurement method to compare each team’s forecast accuracy. DeepMind’s AlphaFold artificial intelligence scored 90 points, which is comparable to the results predicted by the lab.
AI learning speed is astonishing
Image source,fake images
The single spike protein of the new coronavirus (red part)
In the latest round of prediction results (Casp-14), the accuracy of two-thirds of the protein shape predicted by AlphaFold was consistent with laboratory results. Although some other prediction results scored high, the precision did not reach the same level.
AlphaFold uses deep learning algorithms to learn and study the three-dimensional shapes of known proteins stored in global databases. The structure of these folded proteins is presented in a spatial diagram.
BBC Science Reporter Helen Briggs said the speed of learning AI is staggering and it takes a few days to reach the level of research in the laboratory for decades.
Use and meaning
Determining the three-dimensional structure of proteins is essential for the development of new drugs and the understanding of cancer, dementia, and infectious diseases.
Taking the new corona as an example, scientists have been trying to study how the spike protein on the surface of the new corona virus interacts with receptors on human cells.
Professor Martin from University College London told the BBC reporter that understanding how protein sequences fold into three-dimensional shapes is actually one of the most basic problems in biology. He explained that the function of the protein depends entirely on its three-dimensional structure and shape, and the function of the protein is related to everything related to our health and disease.
Therefore, understanding the three-dimensional structure of proteins helps people to design new drugs and prevent and treat diseases, whether they are genetic diseases or infectious diseases.
One of the greatest mysteries in biology.
Professor Dame Janet Thornton from the European Institute of Bioinformatics said that the folding of proteins into unique and beautiful three-dimensional structures is one of the greatest mysteries in biology.
He explained that if we can better understand and predict the structure of proteins, it means that humans will better understand life, evolution, disease, and health and other issues.
Next, more scientists hope to test this data to determine how accurate and detailed the AI methods are.
Today, there are still gaps in human knowledge about proteins, including how many proteins combine and how proteins interact with other molecules, such as deoxyribonucleic acid (DNA) and ribonucleic acid (RNA).
Dr. Krestavovich said that the structure of a single protein is now basically resolved. It opens the door to new ways of finding the structure and shape of protein complexes in the future. It is the joint action of these numerous protein complexes that forms the main mechanism of life and other functions.
