Rapid genotype recognition of human adenovirus based on surface-enhanced Raman scattering combined with machine learning.

Currently, more than 100 genotypes of human adenovirus (HAdV) are classified into seven species (A–G). The tissue tropism and virulence of HAdV are closely related to the genotypes, different genotypes of HAdV infection lead to different types and severity of the disease. Although traditional genotype identification techniques are widely used, they are time-consuming, complex and low sensitivity. Therefore, a rapid and accurate genotype recognition method for HAdV is necessary. In this study, silver nanoparticles incubated with iodine ions and aggregated with calcium ions were used as the enhanced substrate to form a good "hot spot" suitable for viral nucleic acid. The method could rapidly identify the seven common genotypes of respiratory HAdV characteristic fingerprints, combined with machine learning technology could classify and identify HAdV (500 copies/mL) within 5 min, with an accuracy rate of over 98%. In addition, the concentration dependence curve assessing the relationship between the intensity of the characteristic peak and the nucleic acid concentration showed a good linear relationship, the method can be used to quantitatively detect HAdV nucleic acid. This technique can enhance diagnosis and genotype identification of HAdV infection, thus providing a feasible method for the efficient response to respiratory human adenovirus outbreaks. • The Ag@ICNPs can be used to capture SERS signals of multiple genotypes of HAdV. • This method can be used to identify genotypes when concentration is 100 copies/mL. • CNN can identify the genotypes of HAdV with an accuracy rate of over 98%. • Compared with the existing genotype recognition method, this is faster and easier. [ABSTRACT FROM AUTHOR]