Novel process for accurate data analysis from NMR spectrometers

Researchers created a process employing machine learning to enhance data examination from a robust scientific tool—nuclear magnetic resonance (NMR). The NMR data can be utilized to get a thorough knowledge of the proteins and chemical reactions in the human body. NMR is linked to magnetic resonance imaging (MRI) for medical diagnosis.

Scientists have taught computers to interpret NMR results, which could improve the accuracy and speed with which NMR spectroscopy is analyzed. Image Credit: Getty Images.

NMR spectrometers help researchers analyze the structure of molecules, like proteins; however, even with greatly skilled human experts, it requires a substantial amount of time to examine the data. The novel machine learning technique created can examine the data swiftly and precisely.

The researchers elaborate the mechanism, which significantly instructs computers to interpret complex data on atomic-scale properties of proteins, analyzing them into individual, readable images. The research was published in the Nature Communications journal.

To be able to use these data, we need to separate them into features from different parts of the molecule and quantify their specific properties. And before this, it was very difficult to use computers to identify these individual features when they overlapped.”

Rafael Brüschweiler, Study Senior Author, Research Scholar, and Professor, Chemistry and Biochemistry, The Ohio State University

The method created by Dawei Li, lead author of the study and a research scientist at Ohio State’s Campus Chemical Instrument Center, instructs computers to scan images from NMR spectrometers. These images, called spectra, show up as hundreds and thousands of peaks and valleys, which, for instance, can reveal changes to proteins or complex metabolite mixtures in a biological sample, like urine or blood, at the atomic level.

The NMR data provide vital information on a protein’s function and significant clues on the happenings in an individual’s body.

However, deconstructing the spectra into readable peaks is mostly complicated as the peaks overlap. The effect is similar to a mountain range, where nearer, larger peaks cloud smaller ones, which might carry vital information.

Think of the QR code readers on your phone: NMR spectra are like a QR code of a molecule—every protein has its own specific ‘QR code’. However, the individual pixels of these ‘QR codes’ can overlap with each other to a significant degree. Your phone would not be able to decipher them. And that is the problem we have had with NMR spectroscopy and that we were able to solve by teaching a computer to accurately read these spectra.”

Rafael Brüschweiler, Study Senior Author, Research Scholar, and Professor, Chemistry and Biochemistry, The Ohio State University

The method involved generating an artificial deep neural network, a multi-layered network of nodes that the computer employs to separate and examine data.

The scientists developed the network and later instructed it to examine NMR spectra by feeding spectra that had previously been examined by a person into the computer and revealing to the computer the earlier known correct result. The process of instructing a computer to examine spectra is similar to teaching a kid to read—the scientists began with very simple spectra.

After the computer was familiar with that, they moved toward more complex sets. Consequently, the researchers fed greatly complex spectra of various proteins and a urine sample from a mouse into the computer.

The scientists noticed that the computer, employing the deep neural network that it was instructed to examine the spectra, could examine the peaks in the highly complex sample with the same precision as a human expert. Besides, the computer carried out the analysis quickly and highly reproducibly.

Employing machine learning as a tool to examine NMR spectra is just one major step in the extensive scientific process of NMR data interpretation remarked Brüschweiler. However, this study optimizes the abilities of NMR spectroscopists, including the users of Ohio State’s new National Gateway Ultrahigh Field NMR Center, a $17.5 million center funded by the National Science Foundation.

The center is anticipated to be commissioned in 2022 and would have the first 1.2 gigahertz NMR spectrometer in North America.