lcpilling.bsky.social
PhD. Genetic epidemiologist
Researching mechanisms of chronic long-term conditions, ageing, and drugs @ University of Exeter, UK
❤Rstats. UCU rep✊. Views my own
He/him. 🏳️🌈ally. 🇬🇧🇪🇺
📷 ShowYourStripes.info
W: https://experts.exeter.ac.uk/19304-luke-pilling
74 posts
1,011 followers
702 following
Prolific Poster
Active Commenter
comment in response to
post
Stender "Reclaiming mendelian randomization from the deluge of papers and misleading findings." doi.org/10.1186/s129...
comment in response to
post
Burgess "Addressing the credibility crisis in Mendelian randomization." doi.org/10.1186/s129...
comment in response to
post
Highlight of developing this release was naming internal function make_dragen_bed() 🐉🛏️ to extract variants from the WGS DRAGEN pVCFs into a BED file 😂
comment in response to
post
Implemented as a function in my R package of useful epidemiology/statistics things
z = c(10,20,30,40,50)
lukesRlib::get_p_extreme(z)
"1.52e-23" "5.51e-89" "9.81e-198" "7.31e-350" "2.16e-545"
github.com/lcpilling/lu...
comment in response to
post
Hi, finxed link: github.com/lcpilling/lu...
RE p-value threshold. Depends how many tests you are doing! Most conservative in Bonferroni (0.05/Ntests) but assumes tests are independent, so I prefer Benjamini-Hochberg adjustment using R function p.adjust() -- though is less conservative
comment in response to
post
Pretty sure virtual attendance is supported!
www.ukbiobank.ac.uk/learn-more-a...
comment in response to
post
Awesome paper
comment in response to
post
What is wrong with you 😧😅
comment in response to
post
I've always thought "immortal time bias" is pretty cool
comment in response to
post
Yes please - I research ageing and geroscience-related themes such as multimorbidity
comment in response to
post
{twistR} for pharmacogenetic epidemiology
TWIST (Triangulation WIthin A STudy) estimates the ‘genetically moderated treatment effect’ - the excess disease or outcomes in individuals prescribed a specific medication that are attributed to genetic variant(s)
lcpilling.github.io/twistR/
comment in response to
post
{gwasRtools} has two main functions for working with GWAS summary statistics:
1. identify loci either distance-based identification, or LD pruning (can use local Plink and reference data)
2. identify nearest gene for given SNP list (e.g., loci from 1)
lcpilling.github.io/gwasRtools/
comment in response to
post
{lukesRlib} contains a mix of functions I find useful in day-to-day epidemiological analysis
- Trivially easy PheWAS analysis with tidy model output
- Quick functions for data manipulation and working with test statistics
lcpilling.github.io/lukesRlib/
comment in response to
post
{ukbrapR} (phonetically: ‘U-K-B-wrapper’) makes UK Biobank cloud platform (RAP) analysis quicker, easier, and more reproducible
In particular, for a given list of diagnostic codes it is trivially easy to ascertain diagnoses from any source (HES, GP, death, cancer, etc)
lcpilling.github.io/ukbrapR/
comment in response to
post
👋
comment in response to
post
Congratulations! Looks completely legit
comment in response to
post
Thank you - really exciting how the human genetics community here is growing
comment in response to
post
Hi! Great list there. I'd be delighted if you could add me, and also some of my amazing colleagues: @drghawkes.bsky.social, @anna123.bsky.social, @carolinefwright.bsky.social, @chundru.bsky.social
comment in response to
post
This was a fantastic session. Remarkable (cardio)metabolic links underpinning all 3 - looking forward to seeing the ongoing trials results over the next few years
comment in response to
post
This is surely one of them. You can buy it in every pub in Middlesbroough, but drive 20 miles in any direction and very unlikely to get one!