Accurate Virus Identification with Interpretable Raman Signatures by Machine Learning

Rapid identification of newly emerging or circulating viruses is an important first step toward managing the public health response to potential outbreaks. A portable virus capture device coupled with label-free Raman Spectroscopy holds the promise of fast detection by rapidly obtaining the Raman signature of a virus followed by a machine learning approach applied to recognize the virus based on its Raman spectrum. In this paper, we present a machine learning analysis on Raman spectra of human and avian viruses. A Convolutional Neural Network (CNN) classifier specifically designed for spectral data achieves very high accuracy for a variety of virus type or subtype identification tasks. In particular, it achieves 99% accuracy for classifying influenza virus type A vs. type B, 96% accuracy for classifying four subtypes of influenza A, 95% accuracy for differentiating enveloped and non-enveloped viruses, and 99% for differentiating avian coronavirus (infectious bronchitis virus, IBV) from other avian viruses. Furthermore, interpretation of neural net responses in the trained CNN model using a full-gradient algorithm highlights Raman spectral ranges that are most important to virus identification. By correlating ML-selected salient Raman ranges with the signature ranges of known biomolecules and chemical functional groups (e.g. amide, amino acid, carboxylic acid) we verify that our ML model effectively recognizes the Raman signatures of proteins, lipids and other vital functional groups present in different viruses and uses a weighted combination of these signatures to identify viruses. The accurate and interpretable machine learning model developed for Raman virus identification presents promising potential in a real-time virus detection system.

Metagenomic characterisation of additional and novel avian viruses from Australian wild ducks

Birds, notably wild ducks, are reservoirs of pathogenic and zoonotic viruses such as influenza viruses and coronaviruses. In the current study, we used metagenomics to detect and characterise avian DNA and RNA viruses from wild Pacific black ducks, Chestnut teals and Grey teals collected at different time points from a single location. We characterised a likely new species of duck aviadenovirus and a novel duck gyrovirus. We also report what, to the best of our knowledge, is the first finding of an avian orthoreovirus from Pacific black ducks and a rotavirus F from Chestnut teals. Other viruses characterised from the samples from these wild ducks belong to the virus families Astroviridae, Caliciviridae and Coronaviridae. Some of the viruses may have potential cross-species transmissibility, while others indicated a wide genetic diversity of duck viruses within a genus. The study also showed evidence of potential transmission of viruses along the East Asian—Australasian Flyway; potentially facilitated by migrating shorebirds. The detection and characterisation of several avian viruses not previously described, and causing asymptomatic but potentially also symptomatic infections suggest the need for more virus surveillance studies for pathogenic and potential zoonotic viruses in wildlife reservoirs.

Adaptation of Influenza Viruses to Human Airway Receptors

Through annual epidemics and global pandemics, influenza A viruses (IAVs) remain a significant threat to human health as the leading cause of severe respiratory disease. Within the last century, four global pandemics have resulted from the introduction of novel IAVs into humans, with components of each originating from avian viruses. IAVs infect many avian species wherein they maintain a diverse natural reservoir, posing a risk to humans through the occasional emergence of novel strains with enhanced zoonotic potential. One natural barrier for transmission of avian IAVs into humans is the specificity of the receptor-binding protein, hemagglutinin (HA), that recognizes sialic acid-containing glycans on host cells. HAs from human IAVs exhibit “human-type” receptor specificity, binding exclusively to glycans on cells lining the human airway where terminal sialic acids are attached in the α2-6 configuration (NeuAcα2-6Gal). In contrast, HAs from avian viruses exhibit specificity for “avian-type” α2-3-linked (NeuAcα2-3Gal) receptors, and thus require adaptive mutations to bind human-type receptors. Since all human IAV pandemics can be traced to avian origins, there remains ever-present concern over emerging IAVs with human-adaptive potential that might lead to the next pandemic. This concern has been brought into focus through emergence of SARS-CoV-2, aligning both scientific and public attention to the threat of novel respiratory viruses from animal sources. In this review we summarize receptor-binding adaptations underlying the emergence of all prior IAV pandemics in humans, maintenance and evolution of human-type receptor specificity in subsequent seasonal IAVs, and potential for future human-type receptor adaptation in novel avian HAs.

Mammalian ANP32A and ANP32B proteins drive alternative avian influenza virus polymerase adaptations

ANP32 proteins, which act as influenza polymerase co-factors, vary between birds and mammals. The well-known mammalian adaptation, PB2-E627K, enables influenza polymerase to use mammalian ANP32 proteins. However, some mammalian-adapted influenza viruses do not harbour this adaptation. Here, we show that alternative PB2 adaptations, Q591R and D701N also allow influenza polymerase to use mammalian ANP32 proteins. PB2-E627K strongly favours use of mammalian ANP32B proteins, whereas D701N shows no such bias. Accordingly, PB2-E627K adaptation emerges in species with strong pro-viral ANP32B proteins, such as humans and mice, while D701N is more commonly seen in isolates from swine, dogs and horses where ANP32A proteins are more strongly pro-viral. In an experimental evolution approach, passage of avian viruses in human cells drives acquisition of PB2-E627K, but not when ANP32B is ablated. The strong pro-viral support of ANP32B for PB2-E627K maps to the LCAR region of ANP32B.

Antiviral Activity and Adaptive Evolution of Avian Tetherins

ABSTRACT Tetherin/BST-2 is an antiviral protein that blocks the release of enveloped viral particles by linking them to the membrane of producing cells. At first, BST-2 genes were described only in humans and other mammals. Recent work identified BST-2 orthologs in nonmammalian vertebrates, including birds. Here, we identify the BST-2 sequence in domestic chicken (Gallus gallus) for the first time and demonstrate its activity against avian sarcoma and leukosis virus (ASLV). We generated a BST-2 knockout in chicken cells and showed that BST-2 is a major determinant of an interferon-induced block of ASLV release. Ectopic expression of chicken BST-2 blocks the release of ASLV in chicken cells and of human immunodeficiency virus type 1 (HIV-1) in human cells. Using metabolic labeling and pulse-chase analysis of HIV-1 Gag proteins, we verified that chicken BST-2 blocks the virus at the release stage. Furthermore, we describe BST-2 orthologs in multiple avian species from 12 avian orders. Previously, some of these species were reported to lack BST-2, highlighting the difficulty of identifying sequences of this extremely variable gene. We analyzed BST-2 genes in the avian orders Galliformes and Passeriformes and showed that they evolve under positive selection. This indicates that avian BST-2 is involved in host-virus evolutionary arms races and suggests that BST-2 antagonists exist in some avian viruses. In summary, we show that chicken BST-2 has the potential to act as a restriction factor against ASLV. Characterizing the interaction of avian BST-2 with avian viruses is important in understanding innate antiviral defenses in birds. IMPORTANCE Birds are important hosts of viruses that have the potential to cause zoonotic infections in humans. However, only a few antiviral genes (called viral restriction factors) have been described in birds, mostly because birds lack counterparts of highly studied mammalian restriction factors. Tetherin/BST-2 is a restriction factor, originally described in humans, that blocks the release of newly formed virus particles from infected cells. Recent work identified BST-2 in nonmammalian vertebrate species, including birds. Here, we report the BST-2 sequence in domestic chicken and describe its antiviral activity against a prototypical avian retrovirus, avian sarcoma and leukosis virus (ASLV). We also identify BST-2 genes in multiple avian species and show that they evolve rapidly in birds, which is an important indication of their relevance for antiviral defense. Analysis of avian BST-2 genes will shed light on defense mechanisms against avian viral pathogens.

Re-evaluation of the evolution of influenza H1 viruses using direct PCA

The history of influenza H1 virus was re-evaluated by applying a new methodology to sequencing data; this objective method enables comparisons among viral types. The approach led to the segregation of all segments of swine and human viruses into three distinct groups: two of them included the pandemic 1977 and 2009 human viruses, and the remaining group may be new in humans. These three groups might have originated from avian viruses and drifted out independently. Genome shifts occurred occasionally among swine viruses; however, distances between avian and swine/human viruses negated the existence of direct shifts from avian viruses. In humans, only one or two viruses appeared each year, which suggests the presence of competition among viruses that migrated freely. All segments drifted continuously under certain rules and constant velocity. Viruses that had caused an outbreak did not appear again over subsequent decades, which may mean populations had become immune to such viruses. In contrast, the viruses in livestock were rather conserved and maintained unique strains in small, separate areas. Such collections of swine strains included human segments, which could become an epidemic in the future.