Virus detection and identification in minutes using single-particle imaging and deep learning

The increasing frequency and magnitude of viral outbreaks in recent decades, epitomized by the current COVID-19 pandemic, has resulted in an urgent need for rapid and sensitive diagnostic methods. Here, we present a methodology for virus detection and identification that uses a convolutional neural network to distinguish between microscopy images of single intact particles of different viruses. Our assay achieves labeling, imaging and virus identification in less than five minutes and does not require any lysis, purification or amplification steps. The trained neural network was able to differentiate SARS-CoV-2 from negative clinical samples, as well as from other common respiratory pathogens such as influenza and seasonal human coronaviruses. Additionally, we were able to differentiate closely related strains of influenza, as well as SARS-CoV-2 variants. Single-particle imaging combined with deep learning therefore offers a promising alternative to traditional viral diagnostic and genomic sequencing methods, and has the potential for significant impact.

The SARS-CoV-2 Spike protein has a broad tropism for mammalian ACE2 proteins

SARS-CoV-2 emerged in late 2019, leading to the COVID-19 pandemic that continues to cause significant global mortality in human populations. Given its sequence similarity to SARS-CoV, as well as related coronaviruses circulating in bats, SARS-CoV-2 is thought to have originated in Chiroptera species in China. However, whether the virus spread directly to humans or through an intermediate host is currently unclear, as is the potential for this virus to infect companion animals, livestock and wildlife that could act as viral reservoirs. Using a combination of surrogate entry assays and live virus we demonstrate that, in addition to human ACE2, the Spike glycoprotein of SARS-CoV-2 has a broad host tropism for mammalian ACE2 receptors, despite divergence in the amino acids at the Spike receptor binding site on these proteins. Of the twenty-two different hosts we investigated, ACE2 proteins from dog, cat and rabbit were the most permissive to SARS-CoV-2, while bat and bird ACE2 proteins were the least efficiently used receptors. The absence of a significant tropism for any of the three genetically distinct bat ACE2 proteins we examined indicates that SARS-CoV-2 receptor usage likely shifted during zoonotic transmission from bats into people, possibly in an intermediate reservoir. Interestingly, while SARS-CoV-2 pseudoparticle entry was inefficient in cells bearing the ACE2 receptor from bats or birds the live virus was still able to enter these cells, albeit with markedly lower efficiency. The apparently broad tropism of SARS-CoV-2 at the point of viral entry confirms the potential risk of infection to a wide range of companion animals, livestock and wildlife.