Thread by @DavidLVBauer

Urgent correction is needed to the many breathless (re)tweets this evening relaying news that antigen Lateral Flow Tests work against "the" Omicron variant

https://twitter.com/CiesekSandra/status/1464925734741368837?s=20

... significant diversity apparent in Nucleocapsid gene (detected by LFTs) in Omicron... 1/n

thanks to urgent sequencing work by @Tuliodna and South African colleagues that was not visible at all just a few days ago.⏩ SO: we need to know the genome sequence of that Omicron case's N gene before drawing conclusions about LFTs efficacy in "the" Omicron variant... 2/n

and not just S-gene dropout / target failure alone. Speaking of which, there are also Omicron variant viruses that DON'T have S-gene dropout, so not clear to what extent this is diagnostic. May be misled more by countries targeting sequencing at S-gene drpouts in coming days. 3/n

Similarly, there is considerable diversity in the Spike RBD -- i.e. the conserved Spike mutations @PeacockFlu cited in the original Pango lineage proposal ( github.com/cov-lineages/p… ) were one tip of the iceberg and not necessarily generalisable. 4/n

case in point, here's Omicron RBD sites 496, 498, 501, and 505 (which appeared all consistently variant in first sequences observed)... so when labs begin to report neutralising antibody titres vs "the" Omicron variant , the question will be exactly WHICH one was tested (!?) 5/n

& similarly, usefulness of that info will be tied to (a) how much the apparent diversity in the (v. early) data reflects the real-world diversity in circulating strains, and (b) how "good" overall our sequencing data / picture overall is.

https://twitter.com/theosanderson/status/1465091632739586056?s=20

Early days. 6/n

n.b. A large part of the uncertainty with LFTs comes in because companies do not disclose exactly which part of the nucleocapsid protein they detect (i.e. the actual antigen). It's not that hard for them to work that out (and they likely know)... 7/n

so, these mutations in Nucleocapsid may not be relevant at all to the LFT antigen tests — or they may be. And of course, many other factors come into play with LFTs (viral loads, viral tropism i.e. where the virus likes to grow physically in your respiratory tract). 8/n

and again, it's hard to make much in the way of judgements in the face of limited sequencing data, esp. when numbers are limited... in a way, we are in the same situation as we watched the emergence of Delta+friends (B.1.617.x). Was not obvious initially B.1.617.2 would 'win' 9/n

, the difference this time is that early assignment of a Greek Letter has made for a much "broader church" of genomic variation, when in fact we may find that only one small "tip" of the current Omicron family tree is problematic. 10/n

and again, the "tree" is only as good as the data we have (and is just a model, not fact!). Still lots of gaps to fill in, e.g. cycles mutation and reversion are pretty unlikely, as in below. 11/n

Better global surveillance would be very useful to solve this. 🤔 *Almost* as a useful as a fully-funded global vaccination programme would be. 12/n

Thanks to @firefoxx66 and Neher Lab for the Nextstrain build, @GISAID, and all working across the world to ensure open, rapid data sharing (I think @Tuliodna needs another thanks here -- SA hasn't exactly got the best ROI the past 12 months for having excellent sequencing. /end