The protein scientist who competed against the computer program says machine learning will advance biotechnology

Vikas Nanda has spent more than two decades studying the complexity of proteins, the highly complex substances found in all living organisms. Scientist Rutgers has long imagined how the unique patterns of the amino acids that make up proteins determine whether they become anything from hemoglobin to collagen, as well as the subsequent and mysterious step of self-assembly in which only a few proteins aggregate to form even more complex substances. .

So when scientists wanted to conduct an experiment that pitted a human with a deep and intuitive understanding of protein design and self-assembly against the predictive capabilities of an artificially intelligent computer program, Nanda, a researcher at the Center for Advanced Biotechnology and Medicine (CABM) at Rutgers was one of those at the top of the list.

Results are now available to see who – or what – could do a better job of predicting which protein sequences would combine best. Nanda, along with researchers from the Argonne National Laboratory in Illinois and colleagues across the country, reports natural chemistry that the battle was intense but decisive. The competition between Nanda and several colleagues against an artificial intelligence (AI) program was won, very slightly, by the computer program.

Scientists have a keen interest in protein self-assembly because they believe a better understanding could help them design a range of innovative products for medical and industrial use, such as artificial human wound tissue and catalysts for new chemicals.

“Despite our extensive experience, AI has performed well or better across multiple datasets, showing the enormous potential of machine learning to overcome human bias,” said Nanda, professor in the Department of Biochemistry and Molecular Biology. of Rutgers Robert Wood Johnson Medical. School.

Proteins are made up of a large number of amino acids joined from end to end. The chains fold to form three-dimensional molecules with complex shapes. The precise shape of each protein, as well as the amino acids it contains, determines what it does. Some researchers, such as Nanda, are dedicated to “protein design”, creating sequences that produce new proteins. Recently, Nanda and a team of researchers designed a synthetic protein that quickly detects VX, a dangerous nerve agent, and could pave the way for new biosensors and treatments.

For largely unknown reasons, proteins self-assemble with other proteins to form important superstructures in biology. Sometimes proteins appear to follow a pattern, such as when they self-assemble into the protective outer shell of a virus, known as a capsid. In other cases, they self-assemble when something goes wrong, forming deadly biological structures associated with diseases as varied as Alzheimer’s and sickle cell anemia.

“Understanding the self-assembly of proteins is critical to making progress in many fields, including medicine and industry,” said Nanda.

In the experiment, Nanda and five other colleagues were given a list of proteins and asked to predict which ones could self-assemble. Their predictions were compared with those made by the computer program.

Human experts, using rules of thumb based on observing the behavior of proteins in experiments, including electric charge patterns and the degree of water aversion, chose 11 proteins that they believe would self-assemble. The computer program, based on an advanced machine learning system, chose nine proteins.

Humans were right about six of the 11 proteins they chose. The computer program scored higher, with six of the nine recommended proteins capable of self-assembly.

Experience has shown that human experts “favor” certain amino acids over others, sometimes leading them to make bad choices. Furthermore, the computer program correctly pointed to some proteins with qualities that did not make them obvious choices for self-assembly, opening the door for further investigation.

The experience has made Nanda, once skeptical of machine learning for protein assembly investigations, more open to the technique.

“We are working to gain a fundamental understanding of the chemical nature of the interactions that lead to self-assembly, so I was concerned that using these programs would lose important insights,” said Nanda. “But what I’m really starting to realize is that machine learning is just one tool among many, like any other. “

Other researchers on the article included Rohit Batra, Henry Chan, Srilok Srinivasan, Harry Fry, and Subramanian Sankaranarayanan, all from Argonne National Laboratory; Troy Loeffler, SLAC National Accelerator Laboratory; Honggang Cui, Johns Hopkins University; Ivan Korendovych, University of Syracuse; Liam Palmer, Northwestern University; and Lee Solomon, George Mason University.

Leave a Reply

Your email address will not be published. Required fields are marked *