Surface-enhanced Raman spectroscopy for the identification of a non-enveloped virus

Direct antigen and antibody detection tests dominate viral diagnostics. However, these types of assays rely on the detection of specific targets, limiting their use in virus discovery or in the diagnosis of infections of unknown etiology. Problems with these methods also arise when a pathogen presents atypically or during instances of coinfection. In this study, we set out to demonstrate that a platform of carbon nanotubes used for virus capture and Raman spectroscopy can be applied to the detection of non-enveloped viruses grown in cell culture. Non-enveloped viruses lack a lipid bilayer surrounding the nucleocapsid. The carbon nanotube platform consists of carbon nanotubes (CNT) grown on a glass substrate with an average intertubular distance similar to the diameter of viruses, creating a matrix into which the virus particles settle upon drop-casting. Raman spectra of the CNT forest onto which viruses were deposited were collected and machine learning models were constructed to differentiate cell culture samples containing JC polyomavirus (JCPyV), a non-enveloped human polyomavirus, from blank culture. The spectra of media containing virus significantly differed from the blank spectra, forming class clusters on a t-distributed Stochastic Neighbor Embedding (t-SNE) plot. Given that there were significant spectral differences between the culture containing virus and the blank culture, this provides evidence that the virus particles are being detected on the CNT forest. The convolutional neural networks trained to identify JCPyV achieved a 5-fold cross-validated classification accuracy of 97%. This further demonstrates the ability of Raman spectroscopy coupled with machine learning to recognize Raman spectral patterns of viruses, including non-enveloped viruses, for identification purposes.