Chat message

Slack / Private Message Drop, pp.767-768 [SLACK_000981] · slack_pm:msg:08037

Date: 2021-05-10 08:54
Type: chat message · slack
recipient: Kristian G. Andersen, Robert F. Garry, Edward C. Holmes
speaker: Andrew Rambaut

Recipients on this medium are inferred from channel membership, not per-message addressing.

t proportion of bat viruses that are carnivore/human transmissible e number of exposures between bats and carnivores (over all locations, and over decades) p(h) probability of human infection per carnivore outbreak p(p) probability of pandemic arising per human infection p_zoonotic = t x e x p(h) x p(p) t is very small but e is very large For the lab scenario N = number of samples collected by the lab p(i) = probability a sample remains infectious p(l) = probability of a lab infection s = number of scientists handling material p_lab = t x N x p(i) x (1 - [1 - p(l)] ^ s) x p(p) In this one, t is very small, all of the p() are < 1 and N is not that big

In context

A single line often inverts meaning once you see what it answers, so neighbouring messages are always shown.

2021-05-10 08:31 Robert F. Garry open

Baric - aka Deep Throat- is on because Po1 got too much press and we're not really coronavirologists.
2021-05-10 08:36 Andrew Rambaut open

Basically you have gazillion bat viruses. A small fraction of these are human/civet/raccoon dog transmissible but you have millions of these and any transmissible viruses will jump. These will then jump to humans. You have numbers and time on your side. The alternative is a lab samples, say, 10,000 bats (all from a few locations). That sample hasto contain the human transmissible one (but it will not be selecting for it, unlike the carnivore reservoir), it has to stay viable as a virus and infect one person (out of a handful that may be working with the samples). Also the one human infectious virus (out of 10,000 has to be the one to stay viable and infect the lab worker.
2021-05-10 08:37 Andrew Rambaut open

If you have zero evidence that SC2 was in the lab then you are left with this vanishingly small prior probability.
2021-05-10 08:44 Robert F. Garry open

? [shared file(s): image.png]
2021-05-10 08:54 Andrew Rambaut

t proportion of bat viruses that are carnivore/human transmissible e number of exposures between bats and carnivores (over all locations, and over decades) p(h) probability of human infection per carnivore outbreak p(p) probability of pandemic arising per human infection p_zoonotic = t x e x p(h) x p(p) t is very small but e is very large For the lab scenario N = number of samples collected by the lab p(i) = probability a sample remains infectious p(l) = probability of a lab infection s = number of scientists handling material p_lab = t x N x p(i) x (1 - [1 - p(l)] ^ s) x p(p) In this one, t is very small, all of the p() are < 1 and N is not that big
2021-05-10 09:05 Robert F. Garry open

Brilliant! Math>>>Figure
2021-05-10 09:10 Andrew Rambaut open

I forgot to add in p(f) = probability that WIV is fibbing
2021-05-10 09:12 Kristian G. Andersen open

Love it @Andrew Rambaut - and yes, this is what we need. For the ones above e >>> N and p(i) is also very low - probably ~1%. p(l) can be calibrated based on previous lab escapes and is also low - much less than 1%, but let's keep it at that. p(h) we can calibrate based on sero surveys, using a one year seroreversion - so 2% ofpopulation in certain areas.
2021-05-10 09:12 Kristian G. Andersen open

According to The Seeker, nobody in their right mind would consider p(f) > 0.... :wink:

Full conversation →