I'm not sure what you mean here? The presence of the world's biggest collection of novel sarbecoviruses in Wuhan doesn't decrease P(Wuhan|natural), but it does

I'm not sure what you mean here? The presence of the world's biggest collection of novel sarbecoviruses in Wuhan doesn't decrease P(Wuhan|natural), but it does increase P(Wuhan|research-related). So it absolutely should change your estimate of P(research-related|Wuhan) in the obvious way, unless your prior for a research-related pandemic is zero. I don't see how that prior could be zero, since the 1977 flu pandemic already killed ~700k people and is uncontroversially accepted to be of research-related origin. The arguments today are much more sophisticated, though. The debate that you linked relied heavily on Pekar's "The molecular epidemiology of multiple zoonotic origins of SARS-CoV-2". It took those conclusions at face value, apparently since nobody on the other side had the mathematical skills to go deeper. But not only is their fundamental approach questionable (why would you trust a model that complex with no opportunity to validate it?), but their implementation was filled with careless mistakes: https://pubpeer.com/publications/3FB983CC74C0A93394568A373167CE#6 https://pubpeer.com/publications/3FB983CC74C0A93394568A37316... https://gillesdemaneuf.medium.com/backstage-story-the-oct-2023-correction-to-pekar-et-al-e167080e957d https://gillesdemaneuf.medium.com/backstage-story-the-oct-20... They partially addressed this in their erratum, but made minimal changes to the language of their conclusions despite the Bayes factors changing by factor of ~10.