select search filters
roundups & rapid reactions
factsheets & briefing notes
before the headlines
Fiona fox's blog

expert reaction to new paper making suggestions about stray dogs being a possible origin of SARS-CoV-2, the virus that causes COVID-19

A study, published in the journal published in Molecular Biology and Evolution, reports on suggestions about stray dogs being a possible origin of SARS-CoV-2, the virus that causes COVID-19.


Professor Paul Digard, Chair of Virology, The Roslin Institute, University of Edinburgh, said:

“Dr Xia’s research takes a very narrow approach to examining the sequence of SARS-CoV-2 for clues to its origin that does not provide compelling support for the hypothesis that dogs were the source of the virus. The research certainly does not justify the press release headline (“…stray dogs as the possible origin of SARS-CoV-2…”) or provide any evidence to that justifies attempting to control the spread of the virus through controlling the dog population (as alluded to by questioners in the Q&A, not Dr Xia).


Dr Elly Gaunt, Sir Henry Dale Research Fellow, The Roslin Institute, University of Edinburgh, said:

“The Xia study cites a low CpG dinucleotide frequency in the SARS-CoV-2 genome as evidence that the virus is adapted to infect dogs, based on a low CpG composition of a distantly related alphacoronavirus that infects dogs. This logic is flawed for several reasons; three key reasons are below. 

1. Canine viruses have similar genomic CpG ratios to human viruses; this low CpG ratio for SARS-CoV-2 is typical for human-infecting viruses and does not indicate an evolutionary origin.

2. TheSARS-CoV-2 CpG ratio is not similar to canine betacoronaviruses. A comparison with alphacoronaviruses is irrelevant as they are from a different evolutionary group.

3. The changes in CpG ratio reported are subtle, and when you compare with other viruses these differences pale into insignificance.

“Firstly, our current best understanding of why RNA viruses suppress CpG in their genomes is that there is a cellular sensor which recognises CpG motifs in  viral transcripts and targets them for degradation; CpG suppression is a means to avoiding this pathway. This CpG suppression is seen across RNA viruses infecting vertebrates, and across their host vertebrate species. There is no evidence that dogs are an outlier in terms of the extent of CpG suppression in their genomes, which one would expect if viruses are under a stronger selective pressure to repress CpG in that particular host.

“Secondly, SARS-CoV-2 is a betacoronavirus, which are a phylogenetically distinct genus of coronaviruses to alphacoronaviruses. Both alpha- and betacoronaviruses have been detected in a range of species including humans and dogs. The study determines that two betacoronaviruses isolated from dogs had CpG ratios of 0.47-0.48, typical for other betacoronaviruses (anything less than 1 indicates CpG suppression), and higher than the genomic CpG ratio for SARS-CoV-2 (0.40). Canine alphacoronaviruses analysed in this study had CpG ratios of 0.37-0.45; SARS-CoV2 falls within that range, but at this scale, that range is quite broad. The other human alphacoronaviruses included in the study (HCoV-NL63; HCoV-229E is curiously absent or unlabelled) had a CpG ratio of 0.41-0.42, very similar to SARS-CoV-2.

“Thirdly, only the CpG composition of coronaviruses are compared in this study, which does not give a sense whether or not the perceived differences in CpG composition are biologically relevant. Rima and McFerran (JGV, 1997) calculated CpG ratios for a range of RNA viruses, and indicated a CpG ratio for HCoV-229E of 0.49 (n.b. there may be differences in the mathematical approaches between the Xia and Rima studies). This is not as low as other RNA viruses (0.41 for bovine diarrhoea virus) or as high as others (0.75 for hepatitis C virus). HIV-2 was not included in this study, but has a CpG ratio of around 0.22. A summary of the CpG ratios of RNA viruses infecting vertebrates (Simmonds et. al., BMC Genomics, 2013, Fig. 6) reports a range of approximately 0.18-0.9; if the values in this paper were plotted on such a scale (to include viruses originating from a diverse range of hosts) no differences would be visible.”


Prof Mick Watson, Personal Chair of Bioinformatics and Computational Biology, Roslin Institute, University of Edinburgh, said:

“Whole genome comparisons show that the most similar sequenced relative to SARS-CoV2 came from a bat.  That does not mean that SARS-CoV2 came from a bat, but it is the best evidence we have right now.

“The theory that SARS-CoV2 originated in dogs seems to stem from speculation about CpG and high ZAP expression.  This is speculative at best and certainly isn’t strong evidence for the link.”


Prof Ben Neuman, Chair of Biological Sciences at Texas A&M University-Texarkana, and Visiting Associate Professor at the University of Reading, said:

“The conclusion that cats or dogs were involved as an intermediate host for SARS-CoV-2 is highly speculative, and should not be presented as fact.  It is going to take some new data to finally solve the mystery of the origin of SARS-CoV-2, but this study is based on re-analysis of old data without any new data.  This study has similarities to an earlier paper that incorrectly claimed a snake origin of SARS-CoV-2 based on analysis of codon usage.

“Coronavirus genomes are all unusual – always very rich in U nucleotides and often a little low in C nucleotides for reasons that remain unclear.  Genome flexibility, ribosome rate control, recognition by nucleoproteins, actions of a U-specific endonuclease and quirks in the coronavirus proofreading system have all been proposed as explanations, without really arriving at any definite conclusion.

“The observation from this paper that SARS-like coronaviruses may have evolved to minimize host CpG-based recognition may have merit.  But at the moment, we know so little about bat and pangolin innate immunity that it would be a mistake to assume that this evolutionary trend was necessarily driven by interaction with an immune system that we coincidentally happen to have already studied, namely cats or dogs.

“I am confident that answers about the origin of SARS-CoV-2 are still out there, somewhere in the natural world, but we have not found them yet.”


Prof James Wood, Head of Department of Veterinary Medicine, and researcher in Infection dynamics and control of diseases, University of Cambridge, said:

“I find it difficult to understand how the author has been able to conclude anything from this study, or to hypothesise much, let alone that the virus causing COVID19 may have evolved through dogs.  There is far too much inference and far too little direct data.  I do not see anything in this paper to support this supposition and am concerned that this paper has been published in this journal.  I do not believe that any dog owners should be concerned as a result of this work.”


‘Extreme genomic CpG deficiency in SARS-CoV-2 and evasion of host antiviral defense’ by Xuhua Xia was published in Molecular Biology and Evolution at 22:00 UK time on Tuesday 14 April 2020.

DOI: 10.1093/molbev/msaa094


Declared interests

None received.

in this section

filter RoundUps by year

search by tag