Monday always means new data at covid19.sanger.ac.uk, but this week also brings new features! Overall picture should come as no surprise: Delta variant growing in proportion and absolute numbers. In fact, next week is likely to be our highest ever count of one lineage🧵
First new feature is "fade areas by uncertainty", or "nebulosity view", as I like to call it. We fade the colouring of local authorities on proportion view to give a visual hint that you shouldn't over-interpret an area with 100% Delta if it only has a handful of sequences.
Related to this, we show confidence intervals on proportion tooltips and in the unstacked version of the proportion graph. Compare the following two local authorities with similar point estimates, but different amounts of data.
Finally a couple of technical points: we fixed the creeping mystery "B" lineages, that are actually Delta, and added support for Pango aliasing so everything goes up to "A" or "B" at the root, and there's no more "other".
The description of genomic analyses in this piece about the spread of the Delta variant in the UK has major inaccuracies, and hindsight bias. 🧵 theguardian.com/commentisfree/…
Early on, the author, @chrischirp, notes that Pakistan and Bangladesh were red-listed on 9 April, but India wasn't until 23 April. There is a reasonable criticism to be made here, but it has nothing to do with variants. Cases and % positivity were already on the rise in India
We know now that delay led to hundreds of importations of the Delta variant. The article says India "first reported a concerning new variant on 24 March...so why didn’t PHE immediately escalate Delta to a variant of concern in March?"
We've pushed the weekly numbers update (thanks @theosanderson!) at covid19.sanger.ac.uk. As expected, in the most recent 2 week window, B.1.617.2 is the most common lineage in England. Still driven by local concentrations of high case numbers (L cases, R B.1.617.2):
A couple of notes on the site. We have updated the text describing ascertainment: we are no longer excluding surge tests, which make up a large fraction of all tests in key areas in recent weeks. I think this provides least biased frequencies now. Feedback welcome.
I've mentioned this before, but please be cautious interpreting proportions where we have few genomes. For example the "100%" in Dover doesn't mean much because it based on 3 genomes in the most recent 2 weeks of data:
It's clear that B.1.617.2 has been growing faster than B.1.1.7 in UK for a few weeks. Key question for policy, not yet fully answered, is how much due to: 1. vaccine efficacy 2. intrinsic transmissibility 3. human epidemiological factors
🧵of my current thoughts on the topic:
Prelim. data show vaccines work somewhat less well against symptomatic infection w/B.1.617.2, so this makes a non-zero contribution to its recent growth. Good threads on vaccine efficacy data from PHE:
But I think magnitude of observed VE reduction, as well as age distribution of B.1.617.2 infections, makes it unlikely to fully explain growth in UK. B.1.617.2 (blue) is a bit older than recent B.1.1.7 (green), but both more similar to each other than B.1.1.7 from December (red)
Last week @PHE_uk designated B.1.617.2 a VOC (excellent summary from @kallmemeg below). Today we released another week's worth of surveillance genomes at covid19.sanger.ac.uk, and I'd like to walk through some of the features & findings. 🧵
First, headline result, as reported by others, is that B.1.617.2 has hit 6% in England in our 2-week rolling average (>10% in the most recent week). Other VOCs & VUIs steady.
It is hard to know exactly what this means for transmissibility. There were hundreds of imported cases in a short span, and while those are not counted here, they arrived in both a quantity and at a moment of loosening restriction that we have not seen before for other variants.
First off: what is the "India variant", well, it turns out that there are actually three clades of B.1.617, which have now been termed B.1.617.1, B.1.617.2 and B.1.617.3. All three clades appeared in India, likely descended from a common ancestor there some time ago.
The "full" VUI that has garnered attention is B.1.617.1, but interestingly B.1.617.2, which does *not* have the E484Q mutation (apparent reversion) is enriched in the most recent sequences in the UK & elsewhere. So the already dumb "double mutant" label makes even less sense now.
A few thoughts on the B.1.617 variant, first seen in India in late 2020, recently seen in >100 cases in the UK, and very much in the news here. TLDR: we should watch carefully, but I don't think any of our best lines of evidence on variants are yet cause for concern. 🧵
In the "variant era", there are 4 kinds of evidence we can use to evaluate a new variant: (1) how fast it is spreading in different places, (2) pre-existing info about specific mutations it carries, (3) lab experiments (ACE2 binding, Ab evasion, etc), (4) real world vaccine data.
3 & 4 are very important (at the end of the day, arguably all that matters is vaccine efficacy), but take time, even when labs around the world are focusing on these questions. I haven't seen anything yet on B.1.617, so won't comment further.