Thread by @jcbarret

One of the key questions about the new variant (B.1.1.7) is whether there is conclusive evidence that it is more transmissible. I don't think we are absolutely certain yet, but I am pretty confident that it is more transmissible. (1/N)

First, what's the alternative hypothesis? A lineage can get "lucky" and increase in frequency because it happened to be present in local circumstances that favour growth (e.g. poor compliance with social distancing). (2/N)

A good example of this is 20A.EU1 that spread widely throughout Europe after the summer (see excellent @firefoxx66 paper: medrxiv.org/content/10.110…). So far, evidence seems to suggest that was just a "lucky" case, rather than a biological change in the virus. (3/N)

But there are three clear differences with B.1.1.7. First, it appeared with 17 coding mutations compared to its most recently observed ancestor, see Figure 2 in virological.org/t/preliminary-…. This is unprecedented in the pandemic: it isn't just any old lineage. (4/N)

Second, it's much harder for a never-before-seen lineage to grow explosively when the pandemic is as active as it has been in the UK in the past few months. It's very different to 20A.EU1's story in the summer, when there was a long quiet period followed by the second wave. 5/N

In other words, if for some reason there was suddenly a favorable environment in London and the SE, it would've favored all lineages, and there were lots that were already very common, which are now being replaced by B.1.1.7. 6/N

Third, several of the mutations in B.1.1.7 have come up in independent analyses looking for possible functional mutations, especially N501Y. For example:

https://twitter.com/RNajmanovich/status/1340614663269195778?s=20

7/N

Items 1 & 3 (tons of mutations; some in parts of the viral genome that are highly suspect) have to change your prior. Item 2, coupled with the observed spread in the UK (which continues with each new sequence we produce) is pretty strong data. 8/N

In addition to all this there are models that try to disentangle cause and effect in the correlation between B.1.1.7 and growth. I think these will be published in some form soon, but even without them, I think the case is strong. 9/9