iiinova — where your digital self keeps evolving

Emil Uffelmann @euffelmann.bsky.social · 21/09/2026

I recently discovered Spotify DJ, which does exactly that I think support.spotify.com/de-en/articl...

support.spotify.com

DJ - Spotify

120

Reposted by Emil Uffelmann

Michel Nivard @michelnivard.bsky.social · 21/09/2026

This is quite useful for genetics, but dependent on whether any of the inputs are genetics specific, might also be useful for psychologists and social scientists. Go from nagelkerke’s R2 to the R2 on the liability scale ( a latent variable that underlies your binary outcome)

162

Emil Uffelmann @euffelmann.bsky.social · 21/09/2026

Thanks! I think it should work for any predictor that behaves like a PGS (that is, normally distributed and linearly related to the liability)

010

Emil Uffelmann @euffelmann.bsky.social · 21/09/2026

📄 Preprint: www.biorxiv.org/content/10.6... w/ Peter Visscher #genetics #statistics #genomics

biorxiv.org

021

Emil Uffelmann @euffelmann.bsky.social · 21/09/2026

Our adaptation of the Lee formula converts Nagelkerke's R² directly to the liability scale, using only population prevalence and sample case fraction. It works on already published estimates, with no individual-level data needed. R function included.

100

Emil Uffelmann @euffelmann.bsky.social · 21/09/2026

The liability-scale R² of Lee et al. (Genet Epidemiol, 2012) makes polygenic score performance comparable, but it takes linear-regression R² as input. Some studies plug in Nagelkerke's R² instead, which we show gives biased estimates.

110

Emil Uffelmann @euffelmann.bsky.social · 21/09/2026

New preprint 🧵 Nagelkerke's R² is widely used to report polygenic score performance for binary traits, but it depends on prevalence and case fraction, so you can't compare it across diseases or studies. We derive a simple conversion to the liability scale. 👇

193

Emil Uffelmann @euffelmann.bsky.social · 21/09/2026

3/ Our adaptation of the Lee formula converts Nagelkerke's R² directly to the liability scale, using only population prevalence and sample case fraction. It works on already published estimates, with no individual-level data needed. R function included.

100

Emil Uffelmann @euffelmann.bsky.social · 21/09/2026

2/ The liability-scale R² of Lee et al. (Genet Epidemiol, 2012) makes polygenic score performance comparable, but it takes linear-regression R² as input. Some studies plug in Nagelkerke's R² instead, which we show gives biased estimates.

100

Reposted by Emil Uffelmann

Jonathan Pritchard @jkpritch.bsky.social · 09/08/2026

Must-know numbers in human genetics -- As many of you know, I'm writing a free online textbook in human genetics. In this blog post I cover a key skill for genome scientists from that book: how to use mental math to figure out key genome properties. jkpritchard.substack.com/p/on-fermi-p...

jkpritchard.substack.com

On Fermi Problems in Human Genetics

and some very useful numbers about human genomes to get you started!

918570

Reposted by Emil Uffelmann

Joni Coleman @jonicoleman.bsky.social · 24/06/2026

I'm at #BGA2026 #BGA26(?) this week. Out of practice with this skeeting thing, so we'll see how we go...! Introductory remarks from the local hosts here in sunny Amsterdam, to be followed by a plenary from the excellent Wouter Peyrot @behaviorgenetic.bsky.social

2124

Reposted by Emil Uffelmann

FinnGen @finngen.bsky.social · 03/06/2026

We are pleased to announce the release of FinnGen DF13 results! 🧬 While the number of participants remains unchanged, DF13 incorporates updated health register data, increasing the number of cases across most disease endpoints. Browsing & download instructions here: www.finngen.fi/en/access_re...

053

Reposted by Emil Uffelmann

Chris Simms @chrisnsimms.bsky.social · 02/06/2026

And here's another, this one for @newscientist.com, on the biggest genome-wide association study yet of Alzheimer’s. It has identified 48 new gene locations associated with the condition, which could help us find drug targets to prevent it. 🧪 #health #medicine www.newscientist.com/article/2528...

newscientist.com

Huge study of Alzheimer’s genetics identifies new drug targets

Almost 50 more genes have been flagged as being linked to Alzheimer’s, along with changes in activity in crucial cells that disappear as dementia progresses

144

Emil Uffelmann @euffelmann.bsky.social · 26/05/2026

🧬 FUMA v2.0.0 is out, updated by Tanya Phung @ CTG Lab New FLAMES module (effector gene prioritization), new QTLs Analysis module, and expanded xQTL datasets in SNP2GENE fuma.ctglab.nl

fuma.ctglab.nl

Functional Mapping and Annotation of Genome-wide association studies

000

Reposted by Emil Uffelmann

Sacha Epskamp @sachaepskamp.bsky.social · 18/04/2026

Exam prep for a research and statistics exam in 2026! sachaepskamp.com/PL2132_games...

14310

Emil Uffelmann @euffelmann.bsky.social · 12/02/2026

To me, it reveals his humility. Not "I know," not "I discovered," but "I think." Shows he had doubts, like any of us.

020

Reposted by Emil Uffelmann

Uku Vainik @ukuvainik.bsky.social · 24/10/2025

I know that Galton called out for sharing data in 1901

1187

Reposted by Emil Uffelmann

Danielle @daniposthu.bsky.social · 14/10/2025

Proud that the third GWAS for Alzheimer's dementia from the PGC-ALZ working group was just posted online! Huge amounts of work, and what a great collaboration! Check out our exciting findings below 👇 @pgcgenetics.bsky.social #ctglab #alzheimer #dementia #GWAS

0104

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

special thanks to Douglas Wightman (shared first author), my PIs @daniposthu.bsky.social & Ole Andreassen, and the whole Alzheimer's disease working group of the Psychiatric Genomics Consortium

000

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

This study was only possible thanks to a collaboration of >100 co-authors, biobanks, pharma partners, direct-to-consumer companies, and, most importantly, the nearly 3 million participants who shared their data. I am immensely grateful for their participation.

100

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

While ~90% of our data were from Europeans, our multi-ancestry design and ancestry-specific summary statistics pave the way for more diverse future AD GWASs.

100

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

We also generated stratified GWAS results by sex, ancestry, and phenotype definition, and versions excluding the UK Biobank. All summary statistics will be made publicly available upon acceptance to maximize reuse and transparency.

120

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

Our polygenic prediction models explained up to 17% and, on average 13% of variance in European cohorts. Given that SNP-heritability estimates are the theoretical upper limit for polygenic prediction, the LDSC estimates are clearly underestimates.

110

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

Interestingly, using LAVA, we estimate that the SNP-heritability of the APOE region alone is ~9%.

100

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

Using SBayesRC, which models a mixture of SNP effect sizes and can better account for large-effect variants, we estimated SNP-heritability at ~19% (vs. 6% from LDSC). Estimates were similar across African and East Asian ancestries.

110

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

AD GWASs have long been plagued by low SNP-heritabilities (~5%), far below the 60–80% twin-based estimates.

100

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

Sncg and Sst, two GABAergic neurons, have been previously shown to be vulnerable early on in the AD disease process, suggesting that AD-associated variants may influence gene expression in vulnerable neuronal subtypes, leading to neuronal cell death.

100

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

We found enrichment for: Upregulated genes in microglia, and downregulated genes in three neuronal subtypes (Sncg, Sst, and L6 IT Car3).

100

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

We also went further: Using differential expression between AD cases and controls, we tested whether up- or down-regulated genes in specific cell types were enriched for genetic signal.

110

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

Previous AD GWASs using gene expression in healthy controls linked genetic risk mainly to microglia. We replicate that finding: genes highly expressed in microglia show strong association with AD risk.

110

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

We identified 118 loci in a multi-ancestry GWAS and 9 more in a European-only GWAS (total = 127 loci). Of these, 48 were novel, including 8 potential drug targets: QPCT, EGFR, KEAP1, SYK, AXL, RRM2B, CACNA1S, and IL23A.

110

Emil Uffelmann @euffelmann.bsky.social · 14/10/2025

Summary: We analyzed ~180K cases & 2.6M controls, identified 127 loci (48 new), improved heritability estimates (19% in Europeans) & PGS prediction (mean 13%), found potential drug targets, and enrichment in microglia and three neuronal cell types. More details below ⬇️

120

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

A big thank you to Wouter Peyrot for his great supervision and teaching me a great deal about stats gen, and to my other co-authors @daniposthu.bsky.social, Alkes Price, as well as to all members of the schizophrenia and major depressive disorder working groups of the psychiatric genomics consortium

000

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

A limiting factor for the usefulness of the BPC approach is the magnitude of R2. While most PGSs explain little variance, some are already proposed to have clinical utility; as GWAS sample sizes increase, their utility will also grow. See the paper for more limitations: rdcu.be/eIjvC

120

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

We show in simulations and empirical data that this simple way of estimating R2 works surprisingly well, outperforming another published approach.

100

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

In a population reference sample (e.g., 1000 Genomes), where the sample disorder prevalence is the same as in the population, the variance of a PGS on the liability scale will be equal to its R2. That is, no phenotype data is required.

110

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

Our approach depends on a valid estimate of R2. Because we wanted to avoid requiring a tuning dataset that will rarely be available in clinical settings, we developed a new way of estimating R2 using GWAS sumstats and public reference data only.

110

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

We also compared the calibration of BPC to other methods using tuning samples (with geno- and phenotype data) and show that it performs similarly at smaller tuning sample sizes, but worse at larger tuning sample sizes. Because tuning samples are difficult to obtain, BPC may often be preferred.

100

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

We note that the calibration of the Pain et al (2022) method can be improved with some simple tweaks, which we explore in the supplement. The BPC approach still achieves better calibration.

100

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

It is also well calibrated in empirical analyses, where we analyzed 9 disorders of varying genetic architectures

100

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

We show in simulations, across different parameter settings, that the BPC approach is very well calibrated, outperforming a published method.

100

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

To evaluate the BPC approach, we use a metric called the Integrated Calibration Index (ICI): the weighted average of the absolute difference between the real disorder probability and the predicted disorder probability, where 0 indicates perfect calibration.

100

Emil Uffelmann @euffelmann.bsky.social · 27/09/2025

This is achieved by transforming a Bayesian PGS (computed using an existing method, e.g., PRS-CS or SBayesR) to its underlying liability scale, estimating the variances of the PGS in cases and controls based on theory, and applying Bayes’ Theorem to compute the probability.

110