Posts

SVG export from DeckGL
Taxonium, my tool for exploring large phylogenetic trees, uses a few tricks to be fast even when exploring trees of 15 million sequences. One is sparsification: there is no need to draw every single node when sometimes thousands are overlapping and occluding those below: where nodes overlap we only draw some of them.
Sep 14, 2023
LabGPT
Yesterday, GPT-4 was released. I had a play to see how well it might be able to plan and run biological experiments. Scroll to the bottom for some caveats. .
Mar 16, 2023
Identifying malaria phenotype papers with GPT-3
Visit papers.phenoplasm.org Some years ago I launched PhenoPlasm which aims to catalogue all reported phenotypes in Plasmodium in a systematic machine- (and human-) readable format. At the time I tried to populate it with all reported phenotypes in Plasmodium falciparum to date, using a series of searches on Google scholar.
Aug 11, 2022
Two tips for experimental design of barseq experiments
In my time I’ve done various bits of analysis for barcode-sequencing (barseq) experiments (in Plasmodium), and sometimes have chats with people who are starting for the first time. Basically these chats tend to often hit the same two beats about experimental design so here’s a quick blog post about them.
Sep 30, 2021
LFD, PCR, PPV, TLA
People are very interested in the positive predictive value of lateral flow tests. That is, if you receive a positive result from the test, how likely is it that you truly are infected?
Apr 9, 2021
Five lines of R
Somebody on Twitter asked me whether B.1.1.7 data from Florida was still compatible with a logistic increase. It’s amazing how simple this sort of thing is to look at with the Tidyverse and nicely formatted SGTF data from Helix.
Mar 30, 2021
Turning a graph back into data
Summary: The ONS Coronavirus Infection Survey is an immensely valuable scientific project measuring prevalence of coronavirus in the community. However some useful data have been released only in graphical form which makes them hard to re-analyse.
Feb 3, 2021
Looking again at Maccabi data for efficacy after initial Pfizer dose
Update(2021-02-03): Maccabi have also now released a different report looking at a cohort, which gives a less rosy picture of initial efficacy. The explanation may be the older age group in the cohort.
Jan 31, 2021
Analysing Maccabi data on initial efficacy of first vaccine dose
This exciting chart has emerged from the Maccabi study, providing evidence for efficacy of the Pfizer vaccine after the first dose. As Eran Segal pointed out on Twitter, the blue line actually includes the green line.
Jan 22, 2021
New-variant compatibility in the ONS infection survey
Code behind this analysis: https://github.com/theosanderson/theo.io/tree/master/content/post/2021-01-22-ons-data The ONS infection survey has come out and there has been a lot of discussion on the apparent decrease in proportion of “new variant compatible” cases.
Last updated on Jan 22, 2021