Thread by @Nrg8000

So
I have been looking at the genomics of Australia's #COVID2019 cases in recent days and I'm confident that there is significant and largely undetected community transmission of cases in Sydney, and there likely has been since Early Feb. Cases possibly total in the 100s
A THREAD

Now first let me say I'm far from an expert in this and most of my knowledge of cladograms comes from bird taxonomy 🙃 so if any journos read this thread please ask actual experts and don't take my word for it. I just want to flag it because it hasn't been commented on much. 2/

But all in all, this is the clade that's spreading in NSW. So far 8/13 cases genomically sampled in NSW are in this clade and share a common ancestor. 3/

It also looks like it has potentially exported cases to New Zealand and Canada. Which you should take as a warning sign of the potential size of this largely undetected outbreak. 4/

Let me explain a bit about what a cladogram shows. Essentially each sample is a taxon, where the branches leading to different samples meet is a node of a hypothetical untested common ancestor. A group of branches and taxa coming from a single common ancestor is called a clade.

So based on this. All these 8 cases in NSW share a common ancestor at the node shown below with the relevant branches, that likely existed prior to Jan 17 (a Wuhan-based sample from that date also shares a common ancestor). This common ancestor is likely from in China.

Trying to establish the date that this clade was introduced to Australia is more difficult. Most of the Australian samples are recent (earlier is Feb 27th) and likely show a cluster that was detected by chance and then had additional cases found as contacts were chased.

Unlike the Washington cluster (analysed by @trvrb bedford.io/blog/ncov-cryp…) where an early case in the clade was sampled on Jan 15th (letting you know that the introduction was prior to then), it is hard to establish with certainty when this clade was introduced to Australia. 8/

However, some info can be inferred from the genetic distance been taxa from this clade. In this diagram the X axis shows # of mutations from the index case, rather than time, and shows that Australia has seen at least 5 degrees (5-9) of mutation within Australia. 9/

Considering that each generation of COVID-19 takes roughly a week and there have been cases five mutations deep it's highly suggestive that this clade has been in Australia largely undetected for at least 5 weeks. 10/

Of course, this is stupidly simplified and the real dynamics are far more complex, but it shows that this almost certainly isn't recent introduction. Luckily @nextstrain distills some of this into time estimates and confidence ranges of last-common-ancestor nodes. 11/

So for example it shows the last common ancestor of the NSW clade as likely existing on Jan 12 (CI 08-01-2020 to 18-01-2020). Keep in mind I think it's more likely that this common ancestor was in China NOT Australia. 12/

But by looking at the LCA estimated date of the two branches that likely WERE in Australia you can get a better idea of the time this clade was likely introduced to Aus. For example has a CI of 08-01-2020 to 03-02-2020, the other has one from 18-01-2020 to 24-02-2020.

This suggests that the clade was imported into Australia by Feb 3rd 2020 and likely a week or so before then (based on the number of mutations) and that it has been spreading since then until it recently got discovered.

I'm going to tag @trvrb so he can hopefully tell me if I've done something horribly wrong in this analysis (again I only have a glancing idea of how to fully read cladograms). But it certainly suggests to me Sydney has a much bigger problem on its hand than it realises. 15/

Estimating numbers here is pretty moot but if each person spreads it to roughly 2.5 other people (without intervention) and it's been going around for about 6 weeks, then 2.5^6 = 244 cases. If that r0 is closer to 3 as Italy suggests then 3^6 = 729 cases. Which would concern 16/

I'll try to revisit this later to look which specific cases are genetically linked (using data from @VirusWatchAU & @abeleung) and get back to y'all but yeah for now I hope this has been illuminating (and concerning). 17/

Oh and also to stop any bigoted losers in their tracks, after 6 generations of transmission this certainly and completely can not be seen as an Asian disease so if you're going to be scared of people on the street please don't discriminate. 18/

@MackayIM as Twitter's resident Aussie virologist, do you have any thoughts on this ^