Reading: Scientists shouldn't rule out lab as source of coronavirus, new study says newsweek.com/scientists-sho… - personally I do not find the analysis in this new paper remotely convincing 1/n
Summary of paper: they infer many things about possible evolution of SARS-CoV-2 (the new virus) from genome analysis and from comparison to similar analyses done on SARS-CoV (the original SARS) see preprint biorxiv.org/content/10.110… 2/n
They do some detailed analyses of genomic diversity, evolution, substitution patterns and more, some of which seems interesting, certainly. However, I am concerned with both some of the methods used and more importantly how they make inferences from the comparisons they did 3/n
First issue: they only compare SARS-CoV-2 to SARS-CoV. It would have been better to ground truth approach & analysis methods on other viruses for which there is more agreement and information on partners of evolution - I get why SARS-CoV is a good comparison but not enough 4/n
Second issue: the make inferences based on looking at the "phases" of epidemics of SARS-CoV-2/SARS-CoV & claim that the genomics diversity / evolutionary pattern is very different in early phase of SARS-CoV-2 vs. early phase of SARS-CoV. 5/n
They also make many inferences based on claim that SARS-CoV-2 patterns resemble what is seen in later SARS-CoV epidemic; and then they conclude that they SARS-CoV-2 patterns suggest it had been evolving in a human host long before it showed up in Wuhan market 6/n
The problem with making such inferences is that this assumes that pattens in genomic diversity, etc of SARS-CoV-2 that correlate some patterns in SARS-CoV (1) have the same causal explanations and (2) are due to the causal explanations that they think happened in SARS-CoV 7/n
Such inferences seem unjustified. Yes, there may be interesting differences in genomic diversity patterns vs. epidemic stage but they did not test enough alternative plausible hypotheses nor show that anyone has ever made similar inferences using this approach in past 8/n
The one alternative hypothesis report testing is the occurrence of a "bottleneck event similar to the late epidemic SARS-CoV" and the evidence against this is that "SARS-CoV-2 appeared without peer in late 2019" does not to me sufficiently counter this hypothesis 9/n
A third issue is the conclusion they have made that SARS-CoV-2 is somehow very well adapted to humans and that it is evolving in ways like a "late stage" epidemic. The basis for this again is in part comparison solely SARS-CoV, which is insufficient. 10/n
They also study substitutions in CoV2 vs. CoV & infer that little adaptive evolution occurred in CoV2 since epidemic started. Based on this & claim that CoV2 is so well adapted to humans they conclude CoV2 potentially was evolving in humans long before Wuhan market spread 11/n
I find this conclusion to be unconvincing for many reasons including: 1) They have not shown their methods work for detecting these patterns 2) Only compare to CoV 3) Not enough testing of alternative hypothesis 4) No strong evidence CoV2 well adapted to humans from start 12/n
I note - there are some interesting analyses in this preprint but I am really just not convinced that they have shown that they can make the inferences they are making based on these analysis 13/n
Following up on this a bit in response to some questions including those from one of the authors of the preprint @ayjchan eg 14/n
One of my concerns w/ conclusions about selection & preadaptation to humans is that the low amounts of variation in SARS-CoV-2 data right now make it difficult to infer detailed information about selection (also see ) 15/n
Another concern is that even if SARS-CoV-2 is really well adapted to humans (as @ayjchan suggests many say this ) this does not mean it has to have been passaged through humans previously. e.g., some invasive species do really well in new environments 16/n
In fact, there is a long history of invasive species doing remarkably, almost absurdly well in new environments without seeing that environment previously and this also is seen in a variety of emerging infectious diseases so seems perfectly plausible here too 17/n
Another concern is, as I mentioned in response to another question, it is unfair to use lack of experimental evidence of adaptive mutations in Sars-CoV-2 compared to the evidence in Sars-CoV as informative given how recent the Sars-CoV-2 epidemic is 18/n
Also following up on another question from @Ayjchan regarding which other viruses to study 19/n
I think it would be important to include in any analysis a collection of viruses (or other microbes) for which there are genomes and for which one has information about early vs late stage pandemics and/or selection / adaptation to a new host 20/n
Then one could see if, using your informatics and logical workflow, one can detect cases of selection / adaptation or population / pandemic patterns that others have reported and/or that serve as useful comparisons to this situation 21/n
Just comparing to SARS-CoV seems limited for many reasons & makes it hard to tease apart if diversity / substitution patterns are useful for inferring anything generally about a virus or epidemic and also could serve as positive / negative controls for informatics workflow 22/n
Also see my comments about adaptive vs adept here
I note - I 100% support idea of digging more deeply into the evolutionary history of Sars-CoV-2 and as I said previously I think there is a lot of interesting analysis in the paper by @Ayjchan et al. - I do however think some of the conclusions are not warranted at this time 24/n
