Prediction of the highly transmissible strains of Sars-Cov-2 virus appearance on the territory of Saint- Petersburg using a recurrent neural network

Автор: Kustova D.V., Kirirenko A.N., Martynkevich I.S.

Журнал: Вестник гематологии @bulletin-of-hematology

Рубрика: Оригинальные статьи

Статья в выпуске: 1 т.18, 2022 года.

Бесплатный доступ

In 2020, the World Health Organization (WHO) declared severe acute respiratory syndrome, caused by the SARSCoV- 2 virus, a pandemic. The transmissivity of SARS-CoV-2 is associated with the binding affinity of the virus spike S-protein to the angiotensin converting enzyme receptor 2 (ACE2), which is a key event in the penetration of the virus into the cell. It is no coincidence that most vaccines are aimed at blocking binding the S-protein to ACE2. Predicting the occurrence of strains with mutations leading to an increase of the S-protein/ACE2 complex affinity is an important task for preventing new outbreaks of the epidemic and responding quickly to the appearance of highly contagious strains. Predicting the appearance of highly transmissive strains is an important biomedical task, including hematology, since it is patients with immunodeficiency and those on immunosuppressive therapy who are at the greatest risk of infection and severe COVID-19. In our article, the variability of amino acid sequences of SARS-CoV-2 S-protein obtained from clinical samples in St. Petersburg from March 15, 2020 to June 16, 2021 was analyzed. Based on the analyzed sequences, the emergence of new mutations in the receptor-binding motif was predicted using a recurrent neural network model and their effect on binding to the ACE2 receptor was evaluated. The appearance of mutations in positions L455, R457, N481, T500 and G504 was predicted and it was shown that these mutations enhance the affinity of binding to ACE2 by increasing transmissivity.

Еще

Sars-cov-2, s-protein, angiotensin converting enzyme 2, mutations, molecular docking, neural networks

Короткий адрес: https://sciup.org/170194034

IDR: 170194034

Статья научная