Optimizing predictive models to prioritize viral discovery in zoonotic reservoirs

Abstract

31 51 52 Coronaviruses are a diverse family of positive-sense, single-stranded RNA viruses, found widely 53 in mammals and birds 1 . They have a broad host range, a high mutation rate, and the largest 54 genomes of any RNA viruses, but they have also evolved mechanisms for RNA proofreading and 55repair, which help to mitigate the deleterious effects of a high recombination rate acting over a 56 large genome 2 . Consequently, coronaviruses fit the profile of viruses with high zoonotic potential. 57There are seven human coronaviruses (two in the genus Alphacoronavirus and five in 58Betacoronavirus), of which three are highly pathogenic in humans: SARS-CoV, SARS-CoV-2, and 59MERS-CoV. These three are zoonotic and widely agreed to have evolutionary origins in bats 3-6 . 60 61Our collective experience with both SARS-CoV and MERS-CoV illustrate the difficulty of tracing 62 specific animal hosts of emerging coronaviruses. During the 2002-2003 SARS epidemic, SARS-63 CoV was traced to the masked palm civet (Paguma larvata) 7 , but the ultimate origin remained 64 unknown for several years. Horseshoe bats (family Rhinolophidae: Rhinolophus) were implicated 65 as reservoir hosts in 2005, but their SARS-like viruses were not identical to circulating human 66 strains 4 . Stronger evidence from 2017 placed the most likely evolutionary origin of SARS-CoV in 67 Rhinolophus ferrumequinum or potentially R. sinicus 8 . Presently, there is even less certainty in the 68 origins of MERS-CoV, although spillover to humans occurs relatively often through contact with 69 dromedary camels (Camelus dromedarius). A virus with 100% nucleotide identity in a ~200 base 70 pair region of the polymerase gene was detected in Taphozous bats (family Emballonuridae) in 71 Saudi Arabia 9 ; however, based on spike gene similarity, other sources treat HKU4 virus from 72 Tylonycteris bats (family Vespertilionidae) in China as the closest-related bat virus 10,11 . Several 73 bat coronaviruses have shown close relation to MERS-CoV, with a surprisingly broad geographic 74 distribution from Mexico to China 12,13,14,15 . 75 76 Coronavirus disease 2019 (COVID-19) is caused by severe acute respiratory syndrome 77 coronavirus-2 (SARS-CoV-2), a novel virus with presumed evolutionary origins in bats. Although 78 the earliest cases were linked to a wildlife market, contact tracing was limited, and there has been 79 no definitive identification of the wildlife contact that resulted in spillover nor a true "index case." 80 Two bat viruses are closely related to SARS-CoV-2: RaTG13 bat CoV from Rhinolophus affinis 81 (96% identical overall), and RmYN02 bat CoV from Rhinolophus malayanus (97% identical in one 82 gene but only 61% in the receptor-binding domain and with less overall similarity) 6,16 . The 83 divergence time between these bat viruses and human SARS-CoV-2 has been estimated as 40-50 84 years 17 , suggesting that the main host(s) involved in spillover remain unknown. Evidence of viral 85 recombination in pangolins has been proposed but is unresolved 17 . S...

Urban-adapted mammal species have more known pathogens

“…The question of how to account for research effort in host-pathogen ecology is an enduring one, and many studies use citation counts as we did here (Olival et al 2017;Guy et al 2019;Albery et al 2020;Becker et al 2020a;Gibb et al 2020). The detection of this additional implied urban bias suggests that citations capture only one component of sampling bias, and researchers working on zoonoses should endeavour to identify similar measures of potential bias in the future.…”

“…To derive a measure of study effort, which often explains substantial variation in pathogen diversity (Olival et al 2017;Mollentze & Streicker 2020), we conducted systematic PubMed searches to identify how many publications mentioned a given mammal species, following previous methodology (Becker et al 2020a). Domestication status used a sensu lato definition based on whether a species has ever been partially domesticated, coded as a binary variable.…”

A Roadmap Toward Predicting Species Interaction Networks (Across Space and Time)

“…This enables us to represent the structure of a network, which previously required the 2 dimensions of an adjacency matrix, with a smaller number of dimensions. The position of each node in this lower dimensional space is then treated as a latent measurement corresponding to the role of that species in the network (Becker et al 2020). Species close together in the latent space should interact with similar set of species (Rossberg et al 2006;Rohr et al 2010).…”

“…Species close together in the latent space should interact with similar set of species (Rossberg et al 2006;Rohr et al 2010). However, these models are sensitive to sampling biases as they are limited to species for which there is already interaction data (Becker et al 2020), and as a result a methodological breakthough is needed to extend these models to species for which there is little or no interaction data.…”

“…The interwoven nature of these processes imposes structure on biodiversity data which is invisible when examined only through a single mechanism or at a single scale. In addition to the recent application ensemble models (Becker et al 2020), machine learning methods show promise to take the field in a radically different direction, by finding structure in data, and synthesizing mechanistic models from different learning frameworks (Desjardins-Proulx, ). Here we provide a proof-of-concept to show how machinelearning models can enable unreasonably effective prediction of species interactions, whereby we construct a metaweb of host-parasite interactions across space.…”