Thread by @MichaelWorobey

A follow-up to this thread about SARS-related CoVs in bats in Europe that are just one mutation away from a functional furin cleavage site (FCS).
1/

https://twitter.com/MichaelWorobey/status/1471546513281019905

The authors of that study reported a very low frequency of mutants within the bat samples with mutations that would give them a functional furin cleavage site. @jbloom_lab has emphasized that those data cannot be taken as indicative of the presence of those mutants...
2/

Because they could simply be errors. Although both the authors and I noted this, I agree with Jesse that the sequencing data provide no evidence of the presence of the mutants.
3/

However, @jbloom_lab has gone one step further, and argued that the sequencing results positively prove that the mutants are absent in the samples. The data do not support this conclusion any more than they support the conclusion that the mutations are present.
4/

But the demonstration of viruses in bats just one mutation away from a functional cleavage site, does provide very strong evidence for a natural origin of SARS-CoV-2. And there are some reasons to think that in any given bat with a dominant variant 1 mutation...
5/

...away from a FCS there may be minor variants that code for that FCS. And when we consider all bats infected with these viruses, it is certain that some code for functional FCSs.

Here's why: viral populations are huge. Even within a host they can be huge.
6/

Within an individual human, here is the number of new virions generated per day for some viruses I'm familar with:
HIV-1: 10^10 to 10^11.
HCV: 10^12
HBV: 10^13

That's between a trillion and a quadrillion new virus particles per day.
7/

I do not know how many new virions are generated per day in a SARSr-CoV in a bat - perhaps far fewer than the cases above? - but let's consider how sequence space would be explored if the numbers are in the range above.

CoV's have very high fidelity for RNA viruses...
8/

When replicating their RNA genomes. Let's assume 5x10^-6 mutations per genome per replication.

At the low and of the population range (10^10/day), we would see 5,000 mutations across the roughly 30,000 nucleotides of the SC2 genome.
9/

So we could expect to see a mutant that gives rise to a functional FCS once every 18 days or so (since there are 3 nucleotides a give site could change to, but only one would give rise the the FCS.
10/

At the high end (10^13/day) we would see 5 billion mutations. So every day, we would expect to see about 40,000 mutants arise that had a functional FCS.

Again, perhaps bats harbor much lower CoV populations sizes that this, and I'm happy to be informed...
11/

But this is why I think it is incorrect for @jbloom_lab to be stating that he knows that these particular samples do not contain a low frequency of viruses with functional cleavage sites.
12/

And while these variants may be of *zero* consequence in a bat, they will enjoy a tremendous selective advantage in an intermediate host like a raccoon dogs.
13/

This is how natural selection works. It takes random noise at extremely low frequency generated by mutation and molds it into adaptations like the immune systems that allow us to fend off this virus, or the FCS of the virus that puts our immune system to work.
/end

Respectfully, @jbloom_lab is missing the forest for the trees here. One may quibble about error rates in a particular sequencing assay, but the authors of the study have effectively demonstrated that FCSs must exist in bats.
15/14

The authors should be commended for their contribution, not attacked. And complaints that I "blew it up on twitter" are, in my view, unfortunate and misplaced.
15/14

Actually between 10 billion and 10 trillion.