Sars-related coronavirus community outbreak in China in 2019 (SCCOC2019)

Unconditional base rate

https://zenodo.org/record/4067919#.YLgWJCYRXJU

Sars-related coronavirus community outbreak in Wuhan in 1 year (SCCOW2019) | SCCOC2019

Principled update

Wuhan has 1.3% of China's urban population

Probably a slight overestimate; outbreak is possible in rural areas too, but assumed less likely

Sars-related coronavirus community outbreak in Wuhan in 2019 (SCCOW2019)

Unconditional base rate

https://zenodo.org/record/4067919#.YLgWJCYRXJU

Lab leaks from each BSL-3 could be other pathogens, penalised Demaneuf by factor of 2.5 (there are also 2 other BSL-3s in Wuhan with unclear SARS-CoV activity which Demaneuf didn't factor into calculation, so penalty is reduced)

Low BSL level for coronavirus research at WIV

https://www.rootclaim.com/analysis/What-is-the-source-of-COVID-19-SARS-CoV-2

WIV published GOF research with SARSr-CoVs conducted at BSL-2.

Well-adapted for human-human transmission | SCCOW2019

P(evidence|preceeding)

https://www.rootclaim.com/analysis/What-is-the-source-of-COVID-19-SARS-CoV-2

A lab leak causing a community outbreak is very unlikely to be introduced multiple times, and very very unlikely to cause an outbreak despite being initally unadapted for human-human transmission. SARS, MERS and HIV seem to be introduced to humans multiple times but swine flu may have only been introduced once

The above might be non-representative

Makeshift update

https://science.sciencemag.org/content/325/5937/197.full

MERS was never a pandemic and SARS was contained, while H1N1 transmitted globally. Single introductions might favour uncontained pandemics. However, SARS was also much more deadly than H1N1, which also clearly favoured a pandemic in the latter case. HIV, also an uncontained pandemic and very deadly, apparently had relatively few introductions

Maybe there were actually two spillovers?

Makeshift update

https://zenodo.org/record/6291628#.YhwiEXVBzJU

Argues for two spillover events, give it a 50% chance of being correct

Guess: 5% of lab spillovers have 2 lineages (3% per paper + 10% multiple introductions*20% exactly 2) while 15% of zoonotic spillovers have this feature (wild guess)

SCCOW2019 | Well-adapted

No known backbone | above

P(evidence|preceeding)

Presumption: vast majority of lab experiments use known backbones while many zoonoses are previously unknown

WIV seems to have had unpublished sequences

Makeshift update

WIV is very unlikely to publish a backbone for SARS-CoV2 once community transmission has started if they actually had one, so all we know is that they had some closely related viruses that were unpublished before

And additional previously unreleased sequences

Makeshift update

https://twitter.com/franciscodeasis/status/1329057812815765504

Which they had been actively studying in 2017 and 2018

Makeshift update

https://twitter.com/franciscodeasis/status/1367276964135002114

Not clear how many projects WIV has running at a time

https://twitter.com/franciscodeasis/status/1336491158680297477

Maybe more unpublished viruses than usual at WIV

https://www.documentcloud.org/documents/21066966-defuse-proposal

Actually not so unlikely to see unknown backbones; claims of "180 SARSr-CoV strains" seemingly unpublished

SCCOW2019 | Well-adapted, no known backbone

P(evidence|preceeding)

https://yurideigin.medium.com/lab-made-cov2-genealogy-through-the-lens-of-gain-of-function-research-f96dd7413748

No furin cleavage site found in any virus with more than 40% genetic similarity to SARS2; around 100 such viruses known. However, furin well known to increase pathogenicity. Labs are known to add furin cleavage sites to viruses, but this is presumed to be rare among the various things they do with such viruses. Including defuse: assume the base rate of FCS-bearing viruses in lab:natural is 10:1. Assume equal # of natural SARS-like pandemics come from viruses with and without FCS (FCS is rare but better at creating pandemics; but it is not necessary, given we've had a SARS pandemic without FCS). Then lab SARS pandemics are 5/6 bearing FCS

2022 update: addition of furin cleavage sites specifically described in an ECOHEALTH proposal, who has relationship with WIV. Updating to 1/20 lab viruses of equal pathogenicity

P(evidence|preceeding)

https://twitter.com/ydeigin/status/1399171822072041474

Analysis of selections in recent insertions suggests double CGG is somewhat more likely under lab - BF factor 5. I haven't reviewed it personally, but this was endorsed by LL skeptic. DIscounting slightly (5->4) because I think that there are many paths here and there's a good chance the analysis neglects important ones (as I said, not personally reviewed).

SCCOW2019 | Well-adapted, no known backbone, furin site

RBM highly adapted to binding human ACE2 and not matching RATg13/other relatives

P(evidence|preceeding)

https://yurideigin.medium.com/lab-made-cov2-genealogy-through-the-lens-of-gain-of-function-research-f96dd7413748

Assumption: Replacing the RBD and parts of it in a virus backbone is a common practice in gain of function research and was documented on multiple occasions in WIV. Natural recombination can create this result, but also create other results with the same pathogenicity and not just changing the RBM on the RBD. BF is small due to high chance of misunderstanding or double counting "adapted"

RBM may not be from pangolin

Makeshift update

https://twitter.com/franciscodeasis/status/1360795725731364867

Not clear how this affects assessments, but the purported "pangolin" source could be the result of contamination. If contamination was with Bat CoV, maybe raises the likelihood that WIV had sequence? No actual update

A prominent zoonosis advocate thinks the pangolin origin of the RBM is a key piece of evidence for zoonosis, but I don't understand why so I am not updating towards lab origin on the basis of doubt about the pangolin sample

https://twitter.com/K_G_Andersen/status/1399907165666570241

https://www.sciencedirect.com/science/article/abs/pii/S1567134820303245

SCCOW2019 | Well-adapted, no known backbone, furin site, RBM

WIV cut access to virus database in Sept 2019

P(evidence|preceeding)

https://www.researchgate.net/publication/349073738_An_investigation_into_the_WIV_databases_that_were_taken_offline

WIV virus databases are essential clues about the origin of SARS-CoV2, and the official line about "cyber attacks" is not plausible. It is possible that WIV is engaged in a coverup because they don't know the origin, rather than because they know it was from their lab, or simply out of habit

It wasn't completely inaccessible until Feb 2020, but I basically buy the claim that access was restricted somehow - too hard to believe interest went down so much that no-one accessed it and at the same time interest went up so much that people were trying to "hack into it"

and CCP has blocked access to Mojiang mine, insists on reviewing all work related to SARS-CoV2 origins

Makeshift update

https://twitter.com/franciscodeasis/status/1359205971718000643

This is a highly specific act aimed at restricting access to possibly related viral genomes. It is unlikely that it would be taken under a scattergun coverup out of ordinary paranoia

In addition, CCP has restricted access to a databse of academic journals with some reports containing details of WIV virus sampling trips

Makeshift update

https://www.newsweek.com/exclusive-how-amateur-sleuths-broke-wuhan-lab-story-embarrassed-media-1596958

This is another a specific act of restriction targetting ancestral virus genomes. It indicates that WIV/CCP have a coherent theory of how they might be implicated in a lab leak (whether they are rightly or wrongly implicated)

https://zenodo.org/record/4064067#.YLg5c5pfhhE

RaTG13 is apparently a fecal sample but with ~no bacteria + the sequence contains unusually many telomeres. This sounds quite off to me (I tried to ask someone more knowledgable, but no response). It seems likely that published information about the sample origin is erroneous, but not clear what that indicates WRT covid origin.

https://www.preprints.org/manuscript/202008.0595/v1

RaTG13 does not bind efficiently to r.affinis bat ACE2 (but may infect bats in some other way)

Coverup possibilities 1. some details are obfuscated to cover up known virus origins, 2. details are obfuscated to cover up unknown virus origins 3. Normal behaviour, no particular motivation

I don't think "habitual coverup" has much weight as all of the above would be independent in that case leading to very low weight in the end, which is why the makeshift updates didn't reduce the zoo scenario much past the first two items. Motivated coverup + ignorance of origin is still possible

SCCOW2019 | Well-adapted, no known backbone, furin site, RBM, evidence of cover up

No closely related viruses in natural reservoirs discovered

P(evidence|preceeding)

https://virologyj.biomedcentral.com/articles/10.1186/1743-422X-6-207

RaTG13 is too different from SARS-CoV2 to be a direct ancestor, which is also true of the other related sequences at WIV. Natural ancestors of H1N1 were discovered within 11 months of outbreak (likely much less, given publication delays), SARS a natural reservoir was found just 2 months after the virus was identified, and for MERS very close relatives were found in camels around 14 months after the first outbreak

Discounted; see update below

https://en.wikipedia.org/wiki/Severe_acute_respiratory_syndrome#Identification_of_virus

Zoo probabilities follow from rule of succession: in 2 of 3 examples has a natural reservoir failed to be identified after 16 months

Discounted; see update below

SCCOW2019 | Well-adapted, no known backbone, furin site, RBM, evidence of cover up, no intermediaries found after 18 months

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4073860/

Lab probabilities are: 0.95 if virus is substantially modified then leaked (in which case no natural ancestor could reasonably be expected to exist), 0.25 if virus is not substantially modified then leaked (=zoonosis probability)

Discounted; see update below

https://www.cidrap.umn.edu/news-perspective/2013/08/blood-study-suggests-camels-could-be-mers-cov-carriers

https://onlinelibrary.wiley.com/doi/full/10.1111/cla.12425

Update: more detailed analysis suggests neither dromedaries nor civets were actually "intermediate" hosts of MERS or SARS respectively but were infected by humans. Animal populations that were infected via humans have been identified for SARS2, so it is not unusual in this respect

Market carries ~1/2 the weight of the overall zoo hyp. Proximity to market much more likely under zoo (~20x), but there's a decent chance of invalidating data issues (sampling bias etc)

https://twitter.com/Ayjchan/status/1499801596213014529

Proximity to live animals within market

Makeshift update

Guess 1/20 chance environmental samples cluster on West side wildlife vendor area if it's a random super-spreader event, but possibility of bias also

Extended data S1 of this preprint should give a better idea of association between samples and positives, but unfortunately missing https://www.researchsquare.com/article/rs-1370392/v1

I've since revised this down slightly now we know animal stalls were over sampled

Pneumonia deaths, more detailed weibo data

Excess pneumonia deaths and more detailed examination of weibo data also cluster close to the market. See Peter Miller's work

Weibo data from Dec not mentioned

Makeshift update

Errors with weibo data might be largely uncorrelated with confirmed case data errors

Most possible confounders probably independent, motivated reasoning induced confounding is the big exception, and motivated reasoning is common

Why use weibo data but no mention of Dec? If they think it's good enough for a null it's good enough for a sensitivity check

I think Peter Miller's analysis of weibo data shows that it generally supports the conclusion that cases cluster around the market

Restriction site pattern thingy

Lack of intermediate animal host

3/4 other recent zoonotic pandemics had intermediate hosts quickly identified. These may or may not have been the source of the pandemic. A lab pandemic could also lead to humans infecting "intermediate hosts" (and did at e.g. mink farms), but is probably generally less well adapted to animal infection

Lack of clear evidence from FOIA/intel/etc

ECOHealth knew a substantial amount of what WIV were up to (but probably not everything), and appear to have no knowledge of pandemic responsibility. US intel has many other sources besides, but appears to be relying mainly on the same evidence we have here. Numerous pieces of "intel" released are nonsense. This is more speculative than lack of intermediate host

Evolutionary history, lineages and so forth

There's probably worthwhile evidence here, but I don't know how to approach it yet