Thread by @MichaelWorobey

1: I want to follow up the thread below with some additional clarification of why we hypothesize that there may be no real #SARSCoV2 genomes transitional between lineages A and B.

https://twitter.com/MichaelWorobey/status/1433959976913891330?s=20

2: @daoyu15 has written a thread asserting that we "toss any genomes that don't fit your conclusions away". I'm afraid this is incorrect on multiple counts.

https://twitter.com/Daoyu15/status/1434038118030594048

3: What we show is that many of the putatively transitional genomes bear obvious evidence of being artefacts - probably due to bioinformatic pipelines, rather than sequencing errors per se. (Issues like calling a site with poor coverage to be the base of a reference genome.)

4: For example, consider the transitional genome '451394' in the figure in our preprint. It is from Sichuan, and it shares 4 substitutions with a pure lineage A genome ('451383'), also from Sichuan: C173T, C23893T, G28878A, C29640T.

virological.org/t/evidence-aga…

5: There's pretty much no chance that such a transitional genome is real, because that would imply that lightning had struck the same 4 spots in both the transitional lineage and *independently* in the pure A lineage.

See Fig. 3 and Methods of:

science.org/doi/full/10.11…

6: In that paper we excluded all transitional genomes from British Columbia and other areas, both ones with obvious signs of problems like above, but *also* ones that shared no apparent homoplasies (independent changes in two lineages at the same site).

7: The logic is this: if the same sequencing and bioinformatic pipelines that led to clear errors in cases where we can test for them were used to generate sequences where we can't, it is sensible to assume they have the same problems. It is downright naive to assume they don't.

8: We can't yet be 100% sure that there's not a real transitional genome among the ones with no additional shared substitutions with pure A or pure B genomes.
But:

"These findings cast substantial doubt on the veracity of C/C or T/T intermediate genomes in early 2020."

9: And for clarification, we believe these are not sequencing errors (which are largely, although not completely, independent), but rather bioinformatics errors where we’re dealing with a ton of non-independence. Same pipeline: same problem.

10: Finally, let's deal head-on with the calumny that we're tossing away genomes that don't fit our conclusions.

We don't have conclusions yet. We're trying to do everything we can to be sure our analyses are not garbage-in: garbage-out.

11: My first response to the question of whether lineage A and lineage B might reflect separate zoonoses from animals was to draft a letter with Jesse Bloom arguing that the transitional genome identified in the WHO report offered a clear refutation of that idea.

12: But now I know a more (including that that particular genome has dozens of unique mutations and is not reliable). So, in light of new evidence, I've modified my views.

13: Which will be a very familiar course of events to every decent scientist reading this - if not to those who trade in reflexive accusations of corruption on the part of anyone who disagrees with them.