3 ms·

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

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.

mikeyouse 4y ago
> 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/1534663705060773888 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.
- tripletao 4y ago
  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/1534632834568052736 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.

mikeyouse 4y ago

> 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/1534663705060773888 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.

tripletao 4y ago
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/1534632834568052736 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.