Hacker News new | past | comments | ask | show | jobs | submit login

What's your estimate of the relative prevalence of an FCS among the likely space of lab-created SARS-like coronaviruses vs. among natural, SARS-like coronaviruses? My understanding is that adding an FCS is a very common method of lab gain of function, but rare among such viruses in nature.

I don't think the FCS is determinative, and I agree Wade's article overstates its significance. In a Bayesian analysis, it still seems to me like it points weakly (at least 3x prevalence?) towards lab origin, though.




It evolved at least 5+ times completely independently in nature in coronaviruses (and in other viruses).

The significance is zero. Nature itself knows how to figure it out.

> but rare among such viruses in nature.

This assertion is false. It is common among betacoronaviruses in nature, not rare. That was the point of those two articles.

We have even found two sarbecoviruses now (directly related to SARS-CoV-1 and SARS-CoV-2) in bats in Thailand which have an FCS.

It points literally 0% towards lab origin.


Could you at least make a guess at the relative frequencies? My understanding was that hundreds of animal sarbecovirus strains are known, and of those only the two that you note above had an FCS (and not the rare CGG). On the other hand, I understand that adding an FCS is a common technique for lab gain of function, so that perhaps 5% of genetically-engineered SARS-like viruses would have one. That's how I got my ~3x.

It would be nice if each individual piece of evidence were all or nothing, either perfect evidence of lab origin in itself or perfectly irrelevant. I don't think real evidence usually comes that way, so it seems valuable to me to try to quantify even weak evidence.

And not that Nobel laureates don't have an unfortunate history of incorrect beliefs later in life: but I assume you're aware David Baltimore considers the FCS significant? He doesn't seem to have said anything else obviously crazy (unlike Mullis, Pauling, etc.), at least.


> It points literally 0% towards lab origin.

Even given the CGG codon? Sorry, not buying it. You can argue that its evidentiary value is weaker than the article suggests, but it's not 0%.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: