Using artificial intelligence to accelerate protein structure determination based on nuclear magnetic resonance (NMR)

Like many nuclear magnetic resonance (NMR) community, Roland Riek professor initially interested in the technology because of its interdisciplinary nature.Professor Riek studied physics, but was fascinated by biology, he chose (the mid - 1990) was the only across the two areas of research - nuclear magnetic resonance (NMR).Riek professor at the Swiss federal institute of technology in Zurich K.W uthrich professor team degree when work is a major power, he said: "they were very devoted development in nuclear magnetic resonance (NMR) technology, professional knowledge is breathtaking.At that time I knew, I would like to continue to work in the nuclear magnetic resonance (NMR)."

Professor Riek enthusiasm for nuclear magnetic resonance (NMR) no loss, to this day, he said: "the reason I like nuclear magnetic resonance (NMR), it is because of the versatility of this technology and resolution.We can use NMR research up to 200 amino acid protein structure of atomic interaction...It's amazing.Although nuclear magnetic resonance (NMR) involved in the analysis process is complicated, but I am very happy to accept the content of the analytic expression spectra this challenge."

Professor Riek research breadth is admirable, including sub - Angstrom resolution to solve the so-called "polymorphic structure", to understand the mechanism of protein aggregation in Parkinson's disease, investigation of enzymes involved in cancer development, the use of chemical induced dynamic nuclear polarization (CIDNP) to improve the signal-to-noise ratio of the nuclear magnetic resonance (NMR), kernel magnetic resonance (NMR) technology progress in cells, and consider the gathered crystal as a possible mechanism of the origin of life.

Although professor Riek every day busy guide contains 15 to 20 researchers research team, and as the ETH senior staff to participate in various meetings, he still likes to practice aspects of nuclear magnetic resonance (NMR) technique, he said: "I probably a week can take a look at our mri facilities, set up the experimental and data analysis."

Using nuclear magnetic resonance (NMR) to observe the protein: what's the difference between the films and snapshot?

Riek professor argues that a key advantage of nuclear magnetic resonance (NMR) is that it can provide understanding of proteins at the atomic level behavior: "this is a bit like being able to take a film shows the plant operation, but 10 billion times smaller."

Riek professor pointed out that NMR in stark contrast to other methods, "such as low temperature electron microscope and X-ray crystallography method has its value, of course, but in the way, you just get the snapshot - protein or really was frozen in the matrix, or be fixed in a crystal.This means that we must do a lot of work to reconstruct the whole process."与此相反的是，核磁共振具有独特的蛋白质研究能力，因为蛋白质会在溶液中不断移动，这为观察它们如何工作提供了许多机会：“你现在就可以了解蛋白质如何折叠，如何移动，如何与其他分子结合，这是一个令人难以置信的强大且广泛适用的方法。”

New equipment, new methods and new insights

And this kind of ability is driven by the availability of high resolution instrument, including in June 2020 at eth Zurich installation brooke 1.2 GHz nuclear magnetic resonance (NMR) system.So far, though, professor Riek using the spectrometer time is limited, but he were impressed to see everything, he said: "this instrument of ultra-high resolution really surprised me, although know what kind of content in theory can be expected, but when I first saw the spectrum diagram, still feel is a big surprise."

Professor Riek plan using the new device in two main ways: "first of all, we want to develop new methods in the high field system.We have been in promoting the boundary of the nuclear magnetic resonance (NMR), you never know how this will push the progress.application场景可能不会立即开发出来，但对于核磁共振，你几乎可以保证，在适当的时候，肯定会有一个研究领域从新方法中受益。我认为这往往是这种开放式、‘蓝天’类型研究的方式。

"Second, we also to advance the understanding of biological molecules in cells is very interested in, so as to develop new treatment of neurodegenerative diseases such as Parkinson's disease.Once you know what happens at the molecular level, we can reasonably deal with these problems, and finally developed a new technique to treat them."

Using artificial intelligence to break information bottlenecks

However, with the possibility of these studies and to have one more question - data.Riek professor said: "the progress of the biological nuclear magnetic resonance (NMR) is hampered by the data availability and processing time.Must first consider the availability of data, it is worth noting that in nuclear magnetic resonance (NMR) research of tens of thousands of protein structure, only a tiny fraction of the original data set provided to other researchers.This is the field of nuclear magnetic resonance (NMR) a huge problem has been solved.

But he believes that the second problem, run the time needed for nuclear magnetic resonance experiment and analysis results, can take advantage of the rapid development of artificial intelligence (AI) to solve.

He said: "at present, completely determine the characteristics of the structure of a protein requires six months to years.Complete all the measurement and analysis all the data is very time-consuming, and requires expert constantly judgment."This bottleneck, he thought, hindered the progress of this field, thus to find a solution to this problem is very difficult.The professor said: "it is not easy to speed up access to the process of nuclear magnetic resonance (NMR) data.We need to develop biological medium and to analysis the limitation of samples, and then run by nuclear magnetic resonance (NMR) pulse sequence set time limit.But we can do is to more effectively use of nuclear magnetic resonance instrument on time, simplify the data analysis.

Riek professor, artificial intelligence provides a way to achieve these two goals: "through training algorithm to evaluate the result of a, and then to real-time automatic modification of the experiment, we can save a lot of time, only run the pulse sequence to solve the problem of structure is necessary.And due to the time of instrument is expensive, we will save money in the process."Riek professor in this field of work is still in its early stages, but he believes that once the initial findings are published, the advantages of this method will be widely used.

Nuclear magnetic resonance (NMR) data processing revolution?

So, Riek professor of artificial intelligence to the use of nuclear magnetic resonance (NMR) daily protein structure determination will have what effect?

He is optimistic that "just imagine, if we can generate a protein structure, for example, only need two weeks of data collection and analysis of five hours.Now this may sound far-fetched, but I think we can be in 1 to 2 years to see this reality.If so, I think it will completely change the way of biochemistry work, because it speeds up the study of highly complex biological macromolecules, and let ago unrealistic things possible."

Which previously on basic research with professor Riek associated accidental application point of view: "you can't predict what might appear a new way to progress.In the field of nuclear magnetic resonance (NMR), maybe there is a way to conduct experiments, shows how a single water molecule and protein interactions.If so, it would be useful?May be.But before we try, we don't know.

"I think that want to try new things is part of human nature, even if you can't imagine an application immediately.In the field of science, which means that the first to come up with a method, and then see if it will get you where."

And the future of cooperation: from the protein structure to protein dynamics

Like many researchers in the field of nuclear magnetic resonance (NMR), professor Riek work also relies on the cooperation: "in addition to work closely with brooke's scientists, we also contact with several groups, they from different angles, using different methods of analysis, contribute to our research target.Protein itself is a complex targets, especially once you join the neurodegenerative disease involved in the aggregation and decomposition process.You must cooperate with multiple researchers, use different methods to extract the maximum information."

This is a job he is keen to promote paradigm: "we are currently building a server, enables us to provide partners with nuclear magnetic resonance (NMR) based on artificial intelligence services - upload your nuclear magnetic resonance (NMR) data, you will be automatically generated structure.Its progress smoothly."

With professor Riek team of these and other progress, the future looks very bright, "in the next few years, I think we will see most of the focus in the study of nuclear magnetic resonance (NMR), on the one hand is to determine the structure of many proteins, each year on the other hand, through the dynamic measurement combined with polymorphic protein structure, to study the dynamics of proteins, how to obtain protein movement and the operation of the full image.This is a very exciting prospect."

If you want to learn more about professor Riek research, please visit:https://chab.ethz.ch/forschung/institute-und-laboratorien/LPC.html

To learn more brooke nuclear magnetic resonance (NMR) products information, please visit://www.videcame.com/zh/products-and-solutions/mr/nmr.html

Professor Roland Riek

Swiss federal institute of technology in Zurich chemistry and application of biological sciences professor physicochemical and biological nuclear magnetic resonance (NMR) team leader, has been working in the ETH since 2007.Before this, Riek professor is La Jolla, California, USA, Salk institute for biological studies, director of the nuclear magnetic resonance (NMR) facilities.