Thread by @jbloom_lab

I have posted an updated version of my pre-print describing #SARSCoV2 sequences from the early Wuhan epidemic that were deleted from the Sequence Read Archive. This revision should clarify some key questions people asked about the original version: biorxiv.org/content/10.110… (1/n)

First, I would like to thank @stgoldst who provided a set of good-faith scientific critiques that he posted as @biorxivpreprint comments on the original version: disq.us/p/2hwabcu (2/n)

My revisions address @stgoldst's comments as well as others posted on @biorxivpreprint or e-mailed to me directly. You can read my detailed response to the comments and description of the revisions here (disq.us/p/2hwapg6). In this thread, I summarize key changes. (3/n)

First, there has been speculation about *why* the sequences were deleted. Subsequent to submission of the original pre-print, I received a redacted copy of the deletion request e-mail, which I have added as Figure 6 of the revised manuscript (and also pasted below). (4/n)

In the revised manuscript, I refer to this e-mail & make clear that I can't determine the authors' motives. However, I do note that I could not find any websites with updated data, & that practical consequence of deletion was that no one was aware the data existed. (5/n)

Second, @stgoldst correctly noted mutations were still listed in paper published in Small (onlinelibrary.wiley.com/doi/full/10.10…). I now explain this clearly. However, again no one noticed mutation list in paper, so practical consequence of SRA deletion was to obscure existence of data. (6/n)

Third, @stgoldst pointed out that my prior use of example e-mail from pangolin coronavirus SRA deletion request was confusing, & could make readers incorrectly think there was connection between the deletion requests. This is valid point, so I have removed that example. (7/n)

Fourth, several people were confused by my mention of two theories I thought unlikely (RaTG13 outgroup faking, & 2-market hypothesis). Therefore, I have shortened this paragraph, although I still think 2-market hypothesis unlikely for reason here:

https://twitter.com/trvrb/status/1408080716286414852

(8/n)

Fifth, a number of people have noted new data are not transformative, since others have previously inferred earliest viruses aren't from market (eg, Kumar @sergeilkp et al: academic.oup.com/mbe/advance-ar…). I fully agree, and cited this paper heavily... (9/n)

... Now I emphasize even more that new data informative but not transformative, & support *existing* idea virus didn't originate at market. I think study getting so much attention because people are hungry for *any* data on early Wuhan #SARSCoV2, even if not transformative (10/n)

Sixth, in original pre-print I could *not* recover two deleted runs from Google Cloud. After pre-print, several people discovered archived copies of these runs they downloaded before June 2020. I added analysis of these two runs, but they do not meaningfully change results (11/n)

Seventh, @bblarsen1 pointed out that I was failing to properly mask primer binding sites in original analysis. I have fixed this. It does not change results, but does slightly reduce fractional coverage on recovered sequences. (12/n)

Again, a detailed summary of my revisions is here (disq.us/p/2hwapg6) and GitHub repo that transparently tracks all time-stamped changes to manuscript and code over entire history of project is here: github.com/jbloom/SARS-Co… (13/n)

I thank @stgoldst & everyone who took time to provide critiques. Origins & early spread of #SARSCoV2 such a contentious topic that many people have decided they want no part of it. So everybody still engaging in good-faith scientific discussion of topic has my respect! (14/n)

Finally, as scientists, we need to continue to do our best to avoid political / opinion-based arguments about #SARSCoV2 origins / early spread, and remain focused on following two questions: (1) How can we get more data? (2) How can we better analyze the data we have? (15/n)