Theo Sanderson
banner
theo.io
Theo Sanderson
@theo.io
researching viral pathogen genomics at @lshtm.bsky.social | theo.io / sandersonlab.org
sometimes
November 19, 2025 at 6:44 PM
Huge congratulations!
November 14, 2025 at 1:29 PM
Many countries have exclusively submitted sequences to GISAID
November 7, 2025 at 2:02 PM
Yes, of course Nextstrain would I'm sure be happy to receive the data in any form, and would love an API that would allow them to filter for just incremental sequence updates
November 7, 2025 at 11:50 AM
Reposted by Theo Sanderson
Closing these public analyses puts the world more in the dark and concretely harms surveillance as it becomes more difficult for variant spotters to contribute to global situational awareness. 4/5
November 6, 2025 at 9:46 PM
The PDF attached at the bottom is a good read
November 6, 2025 at 9:53 PM
If you want to measure the variance of your model in a reproducible way you'd set the seed at the top then make the n different splits without resetting it in between, and then run over all of them.
October 22, 2025 at 2:06 PM
“for reproducibility” is indeed the point sometimes? I.e. that someone else can re-run my script and get the same output. That's a good thing, all else being equal
October 22, 2025 at 1:50 PM
My robot vacuum recognises trailing wires and socks and other stuff it would get tangled in with AI image recognition and avoids them
October 14, 2025 at 10:10 AM