Altai State University scientists have developed an innovative way to diagnose infectious diseases using machine learning

13 February 2021 Department of Information and Media Communications
The authoritative Swiss scientific journal Frontiers in Immunology, covering the issues of immunology and telling about the latest achievements in this field, has published an article by scientists from the Russian-American Anti-Cancer Center of Altai State University and the Department of Physicochemical Biology and Biotechnology of the Institute of Biology and Biotechnology of Altai State University under the leadership of Academician of the Russian Academy of Sciences Olga Lavrik.

The work titled "Domain-Scan: Combinatorial Sero-Diagnosis of Infectious Diseases Using Machine Learning" is devoted to a joint study of scientists from Altai State University with colleagues from Tel Aviv University (Israel) and Max Planck Institute (Germany), which made it possible to form a serodiagnostic approach based on the use of phage peptide libraries.

The approach, called Domain-Scan, is based on the use of next generation sequencing (NGS) in combination with machine learning techniques. Thanks to this, scientists were able to simultaneously study the binding of antibodies of blood serum to dozens of epitopes, that is, parts of macromolecules of viruses, for example, such as HIV-1 and HCV (viral hepatitis C). When the developed method was used to classify unknown sera, it not only provided an accurate classification, but also allowed to identify an epitope specificity profile.

“We have signed a cooperation agreement with Tel Aviv University. For several years now we have been successfully cooperating with the laboratory of Professor Gershoni, one of the leading specialists in the field of phage display, and we plan to continue our research work,'' noted the leading researcher of the Russian-American Anti-Cancer Center of Altai State University, Candidate of Biological Sciences Dmitry Shcherbakov. “The approach, we have developed, is truly breakthrough. And the point is not even the fashionable methods of machine learning and new generation sequencing. The fact is that this method allows to simultaneously obtain information about the spectrum of epitopes with which antibodies from the blood sera of patients interact. This method is very similar to the chip methods that our team has been developing together with Stephen Johnston from the University of Arizona, but, in my opinion, it is more open and accessible to more researchers."

The developed experimental-computational Domain-Scan approach is universal and can be adapted to other pathogens, if sufficient training samples are provided.

“This Domain-Scan approach can be used not only for detecting infectious diseases, but also for diagnosing oncological diseases. According to the profile of antibodies that are in a person’s blood, it is possible to detect various diseases applying domain scanning,” concluded D. Shcherbakov.

Printable version