For straightforward problems, @github Copilot really rocks.
Yes, as a Python dev, I could code this up myself in less than a minute
But why spend mental energy on this if it can be auto-generated?
As a beginner, this could have easily taken me a few minutes. Now it's just seconds
And the only unnatural about this example is that I deleted the output generated by copilot before recording.
Everything else is exactly as it was when I did it.
Nothing contrived, very natural usage pattern.
It's hard to think of a better way of solving this problem.
Copilot chose a very pythonic solution.
There's a good chance handwritten solutions by many developers would be less idiomatic and more confusing.
What really made me start loving @code was its @ProjectJupyter integration.
I've always disliked native Jupyter notebooks for their lack of IDE features (bad vim support, code completion, highlighting etc.)
VSC makes it easy to run things in cells by adding #%%
Highly recommended
Due to @tiangolo's amazing #Typer package, it's super easy to turn this into a quick CLI
No way, Github Copilot understands a Regex I wrote and generates an appropriate error message for it 🤯
BQ.1* and XBB have different geographic foci
BQ.1* is mostly in Africa, Europe and North America
XBB in South (East) Asia
3 countries with similar levels worth watching for comparison and potential co-circulation are:
- Japan
- Australia
- South Korea cov-spectrum.org/explore/World/… 1/
BQ.1.1 and XBB have quite a lot of spike differences and seem to have similar growth advantages - this makes them candidates for co-circulation.
We may only be able to know once we see how the variants do in countries that had a wave with the other one. 2/
Here are the mutations that only occur in one of the two variants:
XBB only: S:V83A, S:Y144-, S:H146Q, S:Q183E, S:V213E, S:G339H, S:R346T, S:L368I, S:V445P, S:G446S, S:F486S, S:F490
BQ.1.1 only:
S:H69-, S:V70-, S:V213G, S:G339D, S:K444T, S:L452R, S:F486V 3/
A general point why one should be extremely skeptical about stories proposing a causal relation between a new lineage/variant and particular symptoms (right now diarrhea):
New lineages always start very rare.
BQ.1.1 makes up at most a few % of infections in recent weeks. 1/
The people speculating that a variant may be responsible don't seem to understand this.
They think "there's news of a new variant, maybe that's what's driving X"
This is flawed. New variants at low percentage are neither responsible for case growth (yet) nor new symptoms 2/
Even if a variant made a particular symptom 10x more likely (huge effect, much bigger than what's likely to happen), if the variant is only at 1% of all cases, it would only increase overall prevalence of that symptom by 10%! Far below where anecdata would allow detection. 3/
Ich freue mich, dass verschieden Forschungsgruppen die Auswirkungen verschiedener Virusevolutionsszenarien modelliert haben.
Leider enthält das Papier Aussagen die über Modellierung von Szenarien hinausgeht und die in meinen Augen nicht mehr haltbar sind. 1/
In der Veröffentlichung & zugehöriger Pressemeldung steht, dass es momentan nicht möglich sei zu prognostizieren welches Szenario eintritt.
Das stimmt leider nicht. Während es in der Tat vor 2 Monaten nicht klar war wie groß der Variantenimpact sein würde hat sich dies verändert.
Mittlerweile sind Varianten bekannt geworden, die eher Szenario 2 als 1 entsprechen.
Sowohl der beobachtete Wachstumsvorteil als auch die im Labor gemessene Immunflucht von BQ.1.1 & co sind auf einem Niveau vergleichbar mit den Unterschieden von BA.4/5 gegenüber BA.2. 3/
With 3 more days of data, here's a look at how far various collections of immune escape variants have progressed so far.
I now count S:493 as a key mutation, so what used to be >=5 (Pentagon) is now >=6.
The more RBD mutations, the faster the growth. cov-spectrum.org/explore/World/… 1/
Looking at the data like this, we reduce the complexity behind numerous emerging variants into 4 buckets of increasing escape. This is not perfect as what's in a bucket is not totally the same, but it solves the problem of having to remember BQ.1.1, BA.2.75.2, XBB etc. 2/
If you want to connect this way of looking at things with various lineage names you may have heard of, this is a summary
Level 3 (not shown as "boring") = BA.4/5, BA.2.75 = BA.2 + 3 RBD muts
Level 4 = BA.4/5 + 1 RBD mut (e.g. BA.4.6, BF.7) or BA.2.75 + 1 RBD mut (BA.2.75.5) 3/
The best way to monitor the impact of new variants for will be not to consider individual lineages but a collection of lineages with similar properties.
Each lineage may make up only a fraction of the swarm. But together they are already more prevalent. 1/ cov-spectrum.org/explore/United…
In the example above, for the US, BA.2.75* lineages make up 40%, BQ.1 makes up 30%, and other lineages make up the other 30%.
This is a challenge for dashboards like the CDC's, that only list individual lineages when they are at above 1% individually. 2/
I'll provide 3 queries for increasing amounts of immune escape.
The more immune escape mutations there are, the faster the growth, but as it takes time for mutations to evolve, the more mutated variants are still at lower share.
The variant wave will be driven by a combination 3/
With 11 days more data, it is becoming quite clear that BQ.1.1 will drive a variant wave in Europe and North America before the end of November
Its relative share has kept more than doubling every week
It has taken just 19 days to grow 8-fold from 5 sequences to 200 sequences 1/
A week ago, @TWenseleers estimated a growth advantage of ~14% per day for BQ.1 over BA.5
This may be a slight overestimate. But note this was for BQ.1 not BQ.1.1 which has a few extra % pts
Evidence quite strong that BQ.1.1 will have >10%/day advantage. 2/
For comparison:
Alpha had ~7%/day growth advantage
Delta had ~11%/day over Alpha
BA.1 Omicron had ~20-25%/day over Delta
BA.2 had ~11%/day over BA.1
BA.4/5 had ~11%/day over BA.2
Looks like BQ.1.1 is less drastic than Omicron vs Delta but comparable to Delta, BA.2 & BA.4/5 waves