Biochemist David Baker (Seattle, United States, 61 years old) leads a technological revolution that could change science and medicine forever. To understand its full potential you have to delve deep into a living being: its DNA. This molecule stores all the instructions needed to make proteins using four-letter combinations: ACGT. Proteins are responsible for almost any biological process you can imagine, from a tree growing to a firefly glowing in the dark, thinking, breathing, digesting your food, thanks to at least 20,000 different proteins. Digest and everything in between.

Understanding how proteins take their shape has been one of biology’s most difficult problems for more than half a century. Transcription and translation of DNA within cells produces a linear sequence of amino acids arranged in a single file. But in fractions of a second that string folds upon itself to form three-dimensional structures capable of hugging, cutting, pasting, absorbing, transmitting, producing. Nothing in the resulting two-dimensional DNA or amino acid sequences allows one to predict what shape the final molecule will have; And it is its shape that determines its function. Calculating all possible sizes of a single protein by conventional methods would take more than 13.7 billion years, the age of the universe. And there are billions of different proteins in nature.

Last summer, AlphaFold, an artificial intelligence system developed by DeepMind, a Google-owned company, solved the shapes of nearly all known proteins: more than 200 million. The historic achievement was made possible by deep learning systems. These sets of algorithms mimic the functioning of neurons in the human brain. Although they are still far from matching our brain’s capacity, they are very efficient at finding patterns in huge databases. Thanks to these systems, resolving the size of proteins is now done in minutes instead of years.

David Baker’s lab at the University of Washington (USA) goes a step further. They have developed several open artificial intelligence systems that create proteins that never existed in nature. The RoseTTAfold system and its successors make it possible to design new proteins with amazing functions with unprecedented ease, such as inhibiting all forms of Covid or fighting diseases with no known cause, such as Crohn’s or idiopathic pulmonary fibrosis. His team is perfecting a system of tools to “speak a protein”, that is, describe its function with your voice and provide its full sequence to the computer. It wants to provide only part of a protein and have the system autocomplete it, as if it were a Google search.

Baker, along with his DeepMind colleagues Demis Hassabis and John Jumper, has won the BBVA Foundation Frontiers Award for Biomedicine. In this interview conducted by videoconferencing, he talks about the immense potential of this technology. One of their most attainable goals is to create a nasal spray that blocks the entry of influenza, syncytial virus, coronavirus and other winter respiratory pathogens thanks to artificially engineered proteins.

Ask. You say that this technology will change the world more than the Stone Age or the Industrial Revolution. Because

Answer. Until recently, all the proteins we know of were created by nature over thousands of years of evolution. They were like a dwarf language that was given to us. Until now, what we used to do was take those old proteins and make small modifications to them to achieve new functions. Similarly, people picked up stones and thrashed them; Thus the first tools of the Stone Age were made. Now, for the first time, we can make new proteins from scratch that do exactly what we want them to do. It is a human technology that takes us beyond the possibilities of biology.

P. What would be its applications?

R. The first thing we’re going to look at is the impact on medicine with better and cheaper drugs. About seven years ago we started developing an icosahedron-shaped protein [un poliedro de 20 caras], Its appearance was similar to the envelope of many viruses, but it was completely artificial. My colleague Neil King added the coronavirus receptor-binding domain to it, and it turned out that the molecule caused strong immunity against the actual virus. A few years later, one of our first proteins has already been approved as a Covid vaccine and is used in humans, for example in Korea. We are also looking for proteins that improve cancer treatment and others that are capable of generating solar energy or serving as new materials. The possibilities are almost endless.

For the first time, we can design brand new proteins that do exactly what we want them to do.

P. What is the limitation of this technique?

R. One way to know the limits is to consider growth. Whatever living things are able to do on this planet is because of proteins. And all those proteins were created by pure chance in a random process of mutation and selection. No set plan. Now let’s think about how for the first time people can design new proteins to solve problems of their own free will. The possibilities are far beyond what we can imagine.

P. Their new system can talk to a computer and design proteins on demand. For example, can you ask for a protein that cures Alzheimer’s?

R. We can give a simple description of a problem and the deep learning system will provide a sequence of proteins with those properties. But the system is still not perfect. Once a new molecule has been designed by the computer, it must be made using conventional methods in the laboratory and checked that it has the desired properties. The innovation is that we can now use nature to speed up this move. Once we have the amino acid sequence of our protein, we encode it into a DNA sequence, a synthetic gene, which we then introduce into a bacterium. And this bacteria basically becomes our protein factory. Can we design a protein that can cure Alzheimer’s? The problem is that we do not fully understand the cause of this disease. Yes, we have created molecules that bind to the pathological protein fibrils that characterize it, but we don’t know if they are the cause. So there’s still a long trial and error and that’s the tricky part. The problem of designing proteins has been solved. The challenge is knowing how to formulate the problem we want to solve. We need a molecular hypothesis. And for this you need to understand the origin of the disease.

P. How reliable is this technology?

R. It depends on the problem. For a simple question, the success rate is 75%. It’s so new that we’re still learning. In the case of influenza, we have been able to design and test proteins in a matter of weeks, for example. This can be very useful if you have to react to a new epidemic. But it is still very difficult with more complex problems. For example: Degrading plastics. It is such a widespread issue that we still do not know how to deal with it.

P. One of his goals is to develop a nasal spray that we can use to protect ourselves against multiple respiratory viruses at once. When do you think this will be possible?

R. It depends, as economic factors come into play. This type of drug would not be that profitable for pharmaceutical companies, so it would be necessary to see if any company, government or non-profit organization would want to develop it. This is a very common problem in the field of infectious diseases. But, from a technical point of view, I think this year we will know whether these sprays work against Covid. And if they work, it’s reasonable that they work against other respiratory viruses as well.

If You Are The Criminal, You Don’t Need To Design Proteins By Artificial Intelligence, Genetic Sequences Of 1918 Flu Virus Are Already Available

P. Do you see any danger in this technique?

R. Nature has already perfected systems to cause death and destruction on a far greater scale than humans. Consider the flu of 1918, which was extremely deadly and spread quickly. If you are a criminal, you don’t really need to design proteins by artificial intelligence because you already have the genetic sequence of the mentioned virus, or Ebola, for example, available to you. As with any such powerful technology, we have to make sure it is not misused, but I think the risk is small.

P. Do you see any reason for concern that DeepMind is owned by Google and they are so secretive about their work?

R. There is a big difference between my lab, which is completely open, we receive visitors from all over the world and we share information, and a company like DeepMind, which is completely closed. When DeepMind publicized one of its major advances in this area two years ago, there were many prophetic comments from the scientific community, warning that at this rate only big tech companies would dominate the technology. I think the reason we created RoseTTAfold, a system that is open to all, contributed to DeepMind finally opening their system to the public because I’m sure there were people within the company who wanted to keep them secret and make money from them. used to like. , DeepMind is still very secretive and I think this asymmetry between them and us will come at a cost. Only in the last few days very powerful researchers in the development of new drugs against Alzheimer’s, solar energy systems and cancer have visited here. Having an open system gives you so many more ideas. The advancement of science benefits from the free exchange of information.

you can follow materia In Facebook, Twitter I Instagrampoint here to get our weekly newsletter,

Source link