Luke Jostins-Dean
lukejostins.bsky.social
Luke Jostins-Dean
@lukejostins.bsky.social
Data scientist at Nightingale Health, associate professor at University of Oxford
Reposted by Luke Jostins-Dean
Parquet works really well with R as well. I use it all the time to work with GWAS summary statistics that won't fit into memory (our latest metabolite GWAS tested 95M+ variants in UKBB). Here's a minimal example of how to filter data on the fly while reading into R: gist.github.com/kauralasoo/f...
Filtering parquet files with dplyr
GitHub Gist: instantly share code, notes, and snippets.
gist.github.com
September 3, 2025 at 3:01 PM
TIHYLTTW is just too perfect a name to give up, even if it makes it less marketable. One of my all-time favourite book titles.
August 25, 2025 at 8:11 PM
Significant after correcting for 0.98 tests
August 22, 2025 at 8:05 AM
Both of these guys talk very directly to my heart
August 16, 2025 at 4:26 PM
A true *history of Britain
July 19, 2025 at 8:49 PM
The first seems to be 18th century, marked by Hume and Gibbon treating the legends as myths built upon a real minor local figure. The second, generally believing that Arthur was entirely fictional (or a broad composite of people) is much more recent (our lifetimes).
July 19, 2025 at 8:43 PM
Nicholas Higham has a book that tries to answer that ("King Arthur. Myth-Making and History"). There are basically two stages: when did people stop believing the Arthurian romances were a true historian of Britain, and when did people stop believing Arthur was likely to be a real person at all.
July 19, 2025 at 8:40 PM
It's a great book but I don't recall it having anything in particular about when people stopped believing in the truth of the Arthurian legends.
July 19, 2025 at 8:03 PM
Nice to see evidence we didn't just dream all these loci up.

The link from MAML2 notch signalling in DCs to systemic IL12/23 levels is interesting. Any idea what MAML2 is doing in DCs? I think most of the work to date has been in salivary glands (where MAML2 somatic mutations drive cancer).
July 8, 2025 at 2:06 PM
@bradleyomics.bsky.social great work guys, well done
July 8, 2025 at 9:13 AM
Reposted by Luke Jostins-Dean
To see which of these underpin susceptibility, we colocalised these with IBD GWAS. Remarkably, we nominate effector genes at an enormous 74 (❗) loci where one has not previously been nominated in @OpenTargets. This therefore SUBSTANTIALLY improves on previous efforts. 10/
July 8, 2025 at 8:51 AM
I should probably say, just to be clear, that Nightingale Health Plc does not recommend eating 15 grams of butter on your slice of bread. There are likely to be negative effects on your health if done frequently, and you also just feel sort of oily afterwards.
June 2, 2025 at 1:41 PM