Hacker News Reader

The prior linked post is a gish-gallop of overconfident motivated reasoning.

"Two lineages" in rapid-succession is weird - but no better explained by two zoonoses (that left no hints in any animal population) than two lab-leaks.

Worobey et al's analysis is at the mercy of Chinese data sources, and even if all early cases were exclusively in the market, and even the live-animal section, that doesn't disprove lab origin.

It might slant things slightly against it - but when no other animals, anywhere from Laos to Wuhan, have revealed any of these supposed intermediate forms – that somehow zoonosed twice, only after reaching Wuhan, with a dangerous new viral feature Wuhan researchers had proposed splicing into bat viruses?!? – the slant remains very much in the other direction.

But you don't have to trust me; even the WHO now wants a better investigation of the prematurely-suppressed lab-origin possibility: https://www.washingtonpost.com/world/2022/06/09/who-sago-cov...

There's a field-wide omerta about gain-of-function risks – who wants to ruin their relationship with the NIH, & admit research that's pretty typical across many leaky labs worldwide, may have killed 20 million people (and counting)?

> "Two lineages" in rapid-succession is weird - but no better explained by two zoonoses (that left no hints in any animal population) than two lab-leaks.

I mean, everyone can stop reading here... This is patently false. Multiple Spillover is a readily reproducible phenomenon that perfectly explains the dual lineages.

Again, spell out specifically the route how two distinct lineages would be present at the market under the lab leak scenario.

> It might slant things slightly against it - but when no other animals, anywhere from Laos to Wuhan, have revealed any of these supposed intermediate forms – that somehow zoonosed twice, only after reaching Wuhan

You literally only need one animal shedding virus - it's how literally every other pandemic started. The market was shut down weeks after it was clear that a pandemic was underway and only after that did the authorities start looking for infected animals. Absence of evidence here isn't remotely a substitution for evidence of the absence.

> with a dangerous new viral feature Wuhan researchers had proposed splicing into bat viruses?!? – the slant remains very much in the other direction.

This part is just false and belies your lack of understanding of the science and the direction of causality. The reason scientists proposed studying the FCS is because it was identified a decade+ ago as both dangerous for zoonosis and readily possible with natural mutation.

All of these virologists studying coronaviruses weren't doing it out of the blue, but because everyone knew that it was the single greatest zoonotic risk we faced.

A lab worker who was careless (if not depressed/malicious) could just as easily carry out two variants as one.

One animal shedded... two variants of virus? How did the animal get it, when no other nearby wild or market animals in the (lackluster) investigation around Wuhan, or since elsewhere, have shown candidate SARS-CoV2/precursor viruses?

They proposed adding the FCS to existing viruses where FCS hadn't been seen before – and yes, the motivation was because that was considered a potential natural risk. But that doesn't mean it was a natural inevitability. Isn't the proximity of the proposal, "add FCS to a bat coronavirus to see how much worse it makes it", to the appearance of SARS-CoV2 at least a little suspicious?

They & others have published lots of work that specifically made existing viruses more transmissable, or more severe – often in labs with a history of leaks. This can be done via evolutionary pressures – serial passage forced-evolution – or via explicit engineering of new chimeric viruses.

Coronavirus crossover was theorized to be a major risk, after SARS/MERS. That didn't mean that actually outracing nature to make the hypothesized worse viruses in accelerated lab processes was a good risk-reducing idea!

> One animal shedded... two variants of virus

Maybe, but more likely two different ones out of the many millions of poorly documented animals in the East Asia animal trade. Certainly more plausible than a single lab engineering two very distinct linages for no good reason.

> That didn't mean that actually outracing nature to make the hypothesized worse viruses in accelerated lab processes was a good risk-reducing idea!

Presumably you can link to the papers published about how they did just such work, or evidence it in some other authoritative way since you make the assertion so boldly?

How would two different non-bat animals shed two different (but closely-related) bat viruses, in a market with no bats, and far away from the bat caves where related viruses are documented? And further, manage this with no evidence of any other animals infected with related viruses in that region, nor along the trade routes any such animals/viruses must have followed?

Those were the two luckiest bat-viruses ever!

On the other hand, the research lab across town had collected hundreds of such bat viruses from far away - suspiciously never fully catalogued.

EcoHealth belatedly reported in late 2021 to the NIH, its funder that it had already, prior to May 2019:

• created a new, chimeric SHC014-WIV1 virus that was more virulent in human tissue than the (relatively benign) natural WIV1 virus

• took a natural virus that was already highly deadly, MERS, & added new binding sites

You can read an overview from The Intercept at <https://theintercept.com/2021/10/21/virus-mers-wuhan-experim...>, or a (partially-redacted) suspiciously-late EcoHealth report to the NIH at <https://www.documentcloud.org/documents/21089573-priority-gr...> which describes these experiments.

Note also that per The Intercept, an EcoHealth spokesman had previously denied any such MERS gain-of-function experimentation had occurred - claiming instead it was only "suggested as an alternative and was not undertaken".

So: EcoHealth lied.

What if someone's secret budget, or pre-grant investigational work, had helped WIV do exactly the FCS addition they'd already asked for funding to do? That'd be a sufficient, and straightforward, way to get a SARS-CoV2-like virus from less-human-adapted viruses in nature.

And while this route isn't proven – perhaps because those involved have been suspiciously uncooperative – this hypothetical origin has more supporting evidence than "luckiest bat viruses ever snuck to Wuhan from a thousand miles away, zoonosed in Wuhan twice in rapid succession, leaving no sick animals or people anywhere else, despite a giant search".

Uh, are you aware that many different species have confirmed infections of Covid-19? The US department of agriculture has a list of different species where the virus was found: https://www.aphis.usda.gov/aphis/dashboards/tableau/sars-das... So evidently the virus has many options for intermediary hosts, there's no reason to require that there must be a bat nest directly on top of the market.

Further, note that we've found very similar Covid viruses in bats that are from caves around ~1000 km apart (RshSTT182 and RmYN02) while the individuals of the species have never been observed to move more than 200 km. Clearly virus strains can get very far even if they aren't literally carried all the way there by a single individual.

But of course, you also had your own evidence: the allegedly highly suspicious fact that a group of academics hadn't finished all their old projects before starting new ones. Oh yes, how unsurprising.

Yes, COVID has since spread back to many animals, & viruses can travel far. But they still haven't found the precursors of the Wuhan outbreak's original strains.

It is suspicious that EcoHealth lied about what research they had performed. It is suspicious that some very-dangerous, previously-undisclosed pre-May-2019 experiments were only revealed in a belated report.

> There's a field-wide omerta about gain-of-function risks – who wants to ruin their relationship with the NIH, & admit research that's pretty typical across many leaky labs worldwide, may have killed 20 million people (and counting)?

Claims like this, without evidence, reallt detract from any other point you want to make. Because it seems that your thumb is pretty heavy in the scale, without any evidence.

No heavier than your thumb on the zoonosis scale.

Where have I put my thumb on the zoonosis scale?

When there are massive institutional incentives to proclaim X, it is only wise to discount, somewhat, proclamations of X.

In Vanity Fair's lab-leak theory story, there was an illustrative detail:

*> Inside the NIH, which funded such research, the P3CO ["Potential Pandemic Pathogen Care and Oversight"] framework was largely met with shrugs and eye rolls, said a longtime agency official: “If you ban gain-of-function research, you ban all of virology.” He added, “Ever since the moratorium, everyone’s gone wink-wink and just done gain-of-function research anyway.”

A whole field wants certain research to happen, with minimal limits. They don't want to admit, or even let themselves believe, that their peers' accident killed 20 million people. It's a hard pill to swallow!

That's reading an awful lot into an unattributed quote, which includes a direct refutation attributed quotes right above it.

It's clear that there have been moratoriums on GoF without stopping "all of virology," so whatever was meant by this "wink wink nudge nudge" is really unclear.

Also, in the prior paragraph, the reporter talks about "details of proposed experiments being secret" but that's just standard practice for literally all government funding of science. As if they shouldn't be secret for some reason in this case.

If this the sort of weak sauce, I mean, come on.

Just a bunch of weird insinuations, and having read more, saying there's "omerta" on this seems even less likely.

How about at least a little bit of real meat?

> There were two distinct lineages at the market

No there were two distinct lineages in two separate crossover events, the market was just one of them.

Which is even worse for the lab-leak hypothesis.

They're literally just two SNPs apart, per my reply to the grandparent and link to Jesse Bloom's Twitter. What do you think excludes the possibility that those arose during human-to-human transmission, whether at the market or during cryptic spread before that?

No, we don’t know there were two separate crossover events. The two lineages are more likely to be the result of a minor mutation having occurred during human-human transmission, i.e. subsequent to the initial (single) crossover event.

Two lineages do not preclude lab origin.

And the data collection was absolutely biased to the market.

These "two distinct lineages" are separated by exactly two single-nucleotide polymorphisms (SNPs). For comparison, omicron differs from the original Wuhan virus by more than fifty SNPs. SARS-CoV-2 averages something around a third of an SNP per human-to-human transition.

https://mobile.twitter.com/jbloom_lab/status/153463060722071...

It's overwhelmingly likely that something around two months of cryptic spread occurred before the market super-spreader event, regardless of whether the origin was natural or research-related. That's plenty of time for these "two distinct lineages" to develop. They could also have developed at the market, either in a single transmission with two mutations, or in two transmissions with an intermediate lineage that died out before it could be sampled.

Were you previously aware that the two lineages were just two SNPs apart? It's bizarre to me that the "two lineages" have received so much press as evidence of natural zoonosis, when they could have arisen so easily in human-to-human transmission.

> Were you previously aware.

Yes, of course.

> It's overwhelmingly likely that something around two months of cryptic spread occurred before the market super-spreader event, regardless of whether the origin was natural or research-related. That's plenty of time for these "two distinct lineages" to develop. They could also have developed at the market, either in a single transmission with two mutations, or in two transmissions with an intermediate lineage that died out before it could be sampled.

Two points - if there were months of cryptic spread, that essentially rules out 95% of the lab leak hypotheses and further reduces the importance of the WIV proximity.

Secondly, Twitter is an obnoxious interface but Pekar who wrote the paper highlighting the divergent sequences has a good thread explaining why it’s such an important data point:

https://mobile.twitter.com/jepekar/status/149984326950040780...

Since the multiple lineages were present in the very first patients as compared to say January 2020, it narrows the dates and source of the first virus considerably.

The very first identified cases had two distinct viruses and were all present at the market. That’s not at all what you would see if the virus had been spreading for a longer period (we’d have intermediate samples) or if someone from the WIV went to the market and infected people (they’d all show the common lineage that all other transmission chains show).

I've seen the paper, and I don't understand its reasoning. Maybe that's my lack of domain knowledge; but Jesse Bloom is impeccably credentialed and he seems to disagree too.

I believe it's possible that:

1. The mutations that separate Lineages A and B occurred in non-human animals, and were introduced to humans in two zoonotic jumps. This is Pekar's argument.

2. Or, the mutations that separate Lineages A and B occurred in humans during cryptic spread, at the market or elsewhere.

If the two lineages were separated by a large number of mutations, then option (2) would be very unlikely--the virus couldn't spread in humans for long enough to accumulate that many SNPs without exploding into a large enough cluster with enough sickness and mortality that emergence would have been discovered earlier (and we'd have found evidence of that spread retrospectively in sero-surveillance of stored samples, etc.).

But it's just two SNPs--as Jesse Bloom notes, 5-10% of single human-to-human transmissions show at least that much difference, as will many two-transmission chains for which the intermediate never got sampled (or longer chains for which many intermediates never got sampled; the over-dispersed "superspreader" nature of SARS-CoV-2's transmission means almost all chains die out, while a few explode). So what excludes (2)?

ETA: I've never found anyone able to explain in their own words why "two lineages" definitely implies two zoonoses despite the tiny genetic distance between them. I strongly suspect that most people making that claim are simply repeating it, and aren't able to understand the paper, and that the reason none of us can understand is that it's nonsense. If anyone believes otherwise then I'd love to hear, though.

> I've never found anyone able to explain in their own words why "two lineages" definitely implies two zoonoses despite the tiny genetic distance between them. I strongly suspect that most people making that claim are simply repeating it, and aren't able to understand the paper, and that the reason none of us can understand is that it's nonsense. If anyone believes otherwise then I'd love to hear, though.

Plenty of us understand - even Bloom, who I generally respect (aside from some sloppy work at the beginning of the pandemic) understands but thinks it's slightly less likely than the theory you're implying. I (and many others!) disagree.

Pekar does a lot of complicated testing and statistics to shore up his case, but "in my own words", it all comes down to timing. We know in which direction the genetic drift of SC2 occurs, which lets us know that Lineage A is the closer than Lineage B to the ancestral bat coronavirus. So Lineage A could mutate into Lineage B, but the opposite is very unlikely.

However, the very first cases reported to the Chinese government - e.g. the very first patients actually were infected with Lineage B. It was several days later until patients began showing up with Lineage A infections. They've reconstructed the phylo trees to the extent possible with the genomes from those first patients and they just don't leave any real possibility that B came from A. The timing just doesn't work. Pekar estimates that Lineage B infections precede Lineage A ones by ~7 days.

This is further buoyed by the rate of divergence in both Lineages being almost exactly the same even though the basic characteristics of each virus are identical. For that to be true, Lineage A. would've had to quietly spread without being IDed, would need to quickly mutate to Lineage B in order for the first sicknesses to be IDed as Lineage B, and then greatly slow down the rate of mutation to maintain the same divergence as B.

This doesn't "disprove" that a single infection could have sparked both lineages - though it's extremely unlikely to have caused two viable viruses and left no trail whatsoever before them in the genetics. It's all a matter or probabilities and IMO, this shows it's far more likely to have had two jumps (similar to how SARS1 and MERS spread) than some hidden ancestor that was extremely similar to SC2 but wasn't noticeably pathogenic.

At very least, this greatly narrows the timeframe and the location of the very first infection. If various vendors at the market were infected with viruses of different lineage within a handful of days of each other, it's completely obvious that this was the site of the first spread. That may seem like an anodyne observation but unfortunately there are still a lot of people who dispute even that much.

I think Bloom is missing something here that would further change his odds in favor of the multiple zoonosis;

https://mobile.twitter.com/jbloom_lab/status/153466370506077...

> Agreed & one scenario under "not seeded by lineage B at market" is seeded by lineage A at market which then evolved into lineage B in humans. Alternatively could have been seeded by lineage A elsewhere & superspreading of derived lineage B at market.

Guo found environmental samples of both Lineage A and Lineage B at the market -- so both were present there -- his second hypothetical would then be highly unlikely.

I follow the steps; I just don't understand why they're treated as dispositive evidence of zoonosis. I also don't think you're correctly describing Bloom's position here. On Twitter, he writes:

> But the first sequences reported by Chinese government are lineage B from seafood market. So either outbreak was not seeded by lineage B at market, or something very complicated happened. I suggest the former, Pekar et al suggest the latter.

https://mobile.twitter.com/jbloom_lab/status/153463283456805...

I'm taking this as a statement that he disagrees with Pekar, and believes SARS-CoV-2 probably originated from a single introduction (natural or otherwise) into humans. That seems well beyond "slightly less likely". Am I mistaken?

Pekar's theory makes sense if we believe that sampling in humans has been almost perfectly good, capturing every important lineage. But why would we assume that? Such confidence in the human sampling is particularly odd when Pekar's theory also requires us to believe that sampling in animals has been perfectly bad, since we've still found no animals infected (except those infected by humans).

Assuming that Lineage A evolved into Lineage B in humans, the hidden ancestors don't need to be less pathogenic than SARS-CoV-2; the IFR is relatively low already, and people die of similar respiratory diseases every day. The novel virus wasn't discovered until mortality became alarmingly high, leaving lots of room for undiagnosed deaths from cryptic spread before that.

> Guo found environmental samples of both Lineage A and Lineage B at the market -- so both were present there -- his second hypothetical would then be highly unlikely.

Why does that make his second hypothetical highly unlikely? I don't understand why two zoonotic introductions into the market are more likely than two human introductions into the market. Perhaps you'd expect to see other sparks thrown off from the cryptic human spread; but almost all chains of transmission die out, with highly stochastic spread until enough people get infected for the central limit theorem to take over. So the absence of intermediate lineages doesn't seem too significant to me, especially when the "missing link" (if it even exists; I'll again note the lineages could have formed in a single human-to-human transmission) is literally just a single SNP from either lineage.

I would be much more fascinated to know how two separate lineages both made their first leap to humans at the same market. The odds on THAT occurring are much, MUCH smaller than those of two lineages that had already made the leap to humans (or two branches descended from the first human leap) being found at the same market during the first few months of an epidemic.