We are excited to share GPN-Star, a cost-effective, biologically grounded genomic language modeling framework that achieves state-of-the-art performance across a wide range of variant effect prediction tasks relevant to human genetics.
www.biorxiv.org/content/10.1...
(1/n)

Thanks, Cristian!! 😄😄😄

Our trisomy work is now online! 🚀

Got curious after reading our short PheWAS paper? 👀
Catch me 🌠 & my talk 🎤 this October at #ASHG25 in Boston!

Yes, I am still endlessly curious about sex chromosome aneuploidies - whether it's somatic (X/Y loss) or germline (trisomies)!! 💘

And Figure 1 in the preview summarizes everything so nicely - way better than we authors could!!

Preview: www.cell.com/cell-genomic...
Original research article: www.cell.com/cell-genomic...

And, the figure one you generated is so informative!!

Thanks @struangrant.bsky.social ‬and @agacita.bsky.social for seeing the uniqueness of our work and guiding others to understand it better!! Really appreciate your vision and insight!!

Wow!! Big thanks for the very lovely preview!! 🫶🫶🫶

TOO MANY analyses -> very DIFFERENT feedback on (several of) our submissions - some thought it was too lengthy, others considered the design creative, rigorous.

Perhaps no need to do standard analyses if the data/question itself is not standard...

Thanks to the @finngen.bsky.social team for years of effort 🫶 in making it a unique resource for genetic discovery, and congratulations to FinnGen researchers (including us 😉) for utilizing the resources smartly!!

Hi - Excited to share our two talks at @eshg.bsky.social about @finngen.bsky.social!

Because of ... I cannot join in person, but I will answer your questions online (or write to me)! 🧬

Feiyi & Zhiyu are in Milan now; discuss the BLUP work with them (if you can catch them❤️‍🔥)!

Enjoy the conference!

8/n
Parental T1D PGS is a strong predictor of T1D risk in children.

About parental PGSs of other AIDs - not adding too much to the prediction, given the high accuracy of T1D PGS.

PGS vs family history? High parental PGSs are usually more informative than a diagnosis of T1D in the mother or father.

7/n
People with AIDs are less likely to have children - for many reasons, including fear of transmitting diseases to offspring.

By estimating T1D risk for their children, couples can have informed family planning.

Also, children at high risk can benefit from targeted screening & early prevention.

6/n
With pTDT, we see 2 main patterns for parent AID -> offspring T1D inheritance:

Pattern 1 – Overtransmission of both HLA & non-HLA PGSs for parental T1D, HYPO, RA, SLE.

Pattern 2 – Transmission occurs only for HLA: overtransmission for parental CD & psoriasis, undertransmission for IBD & MS.

5/n
But our original curiosity is about parents & children...

We can run polygenic transmission disequilibrium tests (pTDT) with AID HLA & non-HLA PGS in FinnGen trios!

A reminder - pTDT checks if AID risk variants are passed on more or less to children with T1D compared to what would be expected.

4/n
Now, let's revisit the question of the shared genetic background! 🧬🧬🧬

Overall, genetic analyses are consistent with epidemiological findings. For example, for HYPO, CD, RA, both HLA & non-HLA show positive effects.

However, for MS & IBD, non-HLA has no effect, while HLA has protective effects.

3/n
Genetic correlation is straightforward but remember HLA - largely impacting T1D & other AIDs - is excluded by LDSC.

If genetic correlation is a genome-wide non-HLA measurement, how can we get HLA back into the game?

We built a within-sample HLA PGS with weighted ridge classifiers in FinnGen.

2/n
To what extent do the identified epidemiological associations result from genetics?🧬🧬🧬

Let’s download some summary stats and estimate some genetic correlations (quickly).

It looks not too bad - if we put the plot of epidemiological analysis aside for comparisons.

1/n
Which parental autoimmune diseases (AIDs) are associated with T1D in offspring?

With 14,571 trios having T1D-affected children and 43,713 matched control trios (1:3) in Finnish nationwide registers FinRegistry, we identified 9 AIDs associated with increased risk of T1D in offspring.

Why @finngen.bsky.social & why T1D?

1⃣ Finland has the world’s highest incidence of T1D.

2⃣ National healthcare data has been collected for decades -> disease status available for 2 generations.

3⃣ FinnGen genotyped 1/10 of the population + family-based legacy cohorts -> 1.2K T1D genotyped trios.

Type 1 diabetes (T1D) and other autoimmune diseases often co-occur in families. Leveraging data from 12.6K genotyped trios in @finngen.bsky.social, our work - online today in Cell Genomics - studied the transmission of parental autoimmune diseases on T1D in offspring.

www.cell.com/cell-genomic...

Join us April 8-9 for @geneticssociety.bsky.social Spring Symposium on Cancer Genetics!! 🧬

I will present our work on #mCAs & #solid_tumors in @finngen.bsky.social.

Register link: learning.ashg.org/products/202...

Excited to connect with you virtually months before the Boston annual meeting!! 🫶

Fascinating research & dataset - the N of embryos is so impressive!!

Thanks Jack!!

Thanks Shai for posting our trisomy work!! :)

Thanks @jakphd.bsky.social‬ for the nice introduction about our trisomy work!!! 🫶🫶🫶