"they had sequenced the samples with Sangon Biotech (Shanghai) after extracting the DNA in December 2019 from their samples.... They deleted the three most contaminated samples from the Sequence Read Archive. They do not know why the samples were then "un-deleted."
I'll make a mental note to add "accidentally attributing SARS-CoV-2 sequences to Antarctic metagenome sequences from 2018" to my commentary on the perils of chimeric reads if I'm ever asked to present on that again.
For people who are curious what other samples are being sequenced at the same time as their samples, I recommend fishing through the unmapped (or otherwise unexpected) reads of your own samples.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
I've been looking at all these plots about vaccine effectiveness and hospitalisation, and thinking they're missing the exposure side of things... so I made another plot.
This plot shows how the vaccine reduces both hospitalisations *and* disease.
I've assumed here that the number of exposed unvaccinated people is the same as the number of unvaccinated people with COVID. I could bump that exposure number up a bit, but it wouldn't change the overall visualisation because the proportions of vaccinated people wouldn't change.
There's another niggle in that vaccination & cases are a dynamic event. MOH doesn't have vaccination status in their case demographic CSV file, so I use 2 wks ago as a fudge.
That shouldn't be too much of a problem anyway, because recent numbers dominate for exponential growth.
In short, they're great. My main comments are around needing more emphasis on participant control - making that implicit principle explicit.
"Researchers are expected to learn as well as gather data in research, to collaborate and to give back to the community (eg, through koha and sharing ideas)." [S 5.7] - a good principle to have; I think a minimal sharing protocol needs to be specified in ethics applications.