Virus detection by untargeted proteomics enabled by a spectral library of the human virome built for diagnostics

Viruses are ubiquitous in the environment. Almost all living organisms are known to host certain viruses, including bacteria, fungi, plants and animals. A small proportion of these viruses can in turn infect humans and cause disease. Viral infections are primarily diagnosed by gene or protein detection of viral components using targeted methods such as PCR and immunoassays. Metagenomics (mNGS), on the other hand, enables the untargeted detection of viral genomes, while there is no well-established comparable proteome-based method.

In this study, we aim to introduce proteomics as a complementary approach for the untargeted detection of human-pathogenic viruses. The viral proteomics (vPro) workflow is based on an in-silico derived peptide library covering the human virome in UniProt (331 viruses, 20,386 genomes, 121,977 peptides), which was designed for diagnostic purposes. The reliability of virus identifications is monitored by a confidence score ("vPro score") specifically adapted to proteomics. In combination with high-throughput diaPASEF-based data acquisition, this workflow enables the analysis of up to 60 samples per day with a specificity of > 99.9 % for virus detection, which was determined in an analysis of 221 samples including plasma and swab samples. The sensitivity of this approach for the detection of SARS-CoV-2 in nasopharyngeal swabs was determined to be ct 27 (CI = 95 %). The comparison of mNGS and proteomics for the detection of SARS-CoV-2 in swab samples revealed, that proteomics is already able to provide the same specificity on species-level and quantitative accuracy with respect to qPCR as mNGS at similar or even higher throughput. Currently, the limitation for virus detection using proteomics is the sensitivity especially compared to targeted virus detection methods but already less pronounced compared to mNGS. This study further demonstrates the potential to improve the sensitivity on various stages of the proteomics workflow in the future.

Taken together, this study provides the basis to introduce proteomics as an alternative approach for untargeted detection of human viruses. Further improvements of the sensitivity should enable its wider adoption for high-throughput studies. The data analysis workflow of vPro can further be integrated into large-scale proteome studies of biofluids such as human plasma to explain outliers due to acute infections and to determine the prevalence of persistent infections, such as SARS-CoV-2.