Slack / Private Message Drop — page 133

[Eddie Holmes] So bizarre that such a clarification is needed. Sums up the current mess we are in. [2020-08-10 05:50:34] [Robert Garry] https://www.today.com/video/inside-the-wuhan-lab-at-center-of-coronavirus-controversy-89836101738?playlist=mml snnd_todayarchivesmonday [2020-08-10 06:25:18] [Robert Garry] Not much in this report. [2020-08-10 06:28:13] [Andrew Rambaut] "So you are 100% sure there was no release?" , "Yes, 100%", "Great thanks" [2020-08-10 08:35:05] [Robert Garry] Yeah - This will not help, but will be harmful. This is like an ex-postdoc of mine [I fired her]. To say she had no lab skills did not capture the essence. She had negative lab skills. [2020-08-10 08:42:12] [Kristian Andersen] Yeah, absolutely nothing in this reporting - but then again, what did they expect to find? It's a lab... like so many other labs. "Five virologists familiar with lab protocols found it improbable...". I guess that's us - which is kinda strange. [2020-08-10 10:43:46] [Robert Garry] It make a little more "sense" now - most of their science consultants are "virologists" that are unfamiliar with lab protocols. [2020-08-10 14:25:53] [Eddie Holmes] Yes, pointless. [2020-08-10 17:10:28] [Kristian Andersen] [thread - ID: 2020-08-10 17:10:28] @Andrew Rambaut - reading your Doug paper in a little more detail since the NYT folks asked about it... Question for you - looking at this figure, I largely see overlap in cluster sizes _earlier_ in the epidemic. Since Dougless came in early and Doug came in later - where, importantly there was A LOT more ongoing transmission - is it possible that when you measure "cluster sizes" (or any of the downstream phylodynamic estimates) that it's possible that a larger subset of Doug clusters that you assign to transmission within the UK could be due to multiple introductions as compared to Dougless? Reading the methods, I'm wondering if you have multiple examples of, say, a cluster actually being made up of two introductions of identical viruses from neighboring countries, but then gets counted as a single introduction on your tree? I assume this effect would be more significant for later clusters - before mitigation - and could therefore skew cluster size towards Doug's being larger because they're more likely to be due to multiple introductions of identical viruses than Dougless'. One way to control for this would be to either make sure that clusters are all from the same city/region - but I assume you run out of power if you start dividing the dataset like that. Separate to a potential skew due to time, there's also a potential skew due to source. If you look at introductions from the same time period, Dougless would primarily be coming from China (less travel), while Doug would be coming from neighboring countries - again, this'd increase the likelihood of multiple introductions of identical Doug