Burak Yelmen

Interpreting artificial neural networks to detect genome-wide association signals for complex traits Abstract. Investigating the genetic architecture of complex diseases is challenging due to the multifactorial interplay of genomic and environmental influe

Our paper on detecting GWA signals from neural networks is now published. Here we propose a model- and score-agnostic post hoc framework to detect potentially associated loci from neural networks trained for phenotype prediction.
academic.oup.com/nargab/artic...

Hello all, we are looking for PhD candidates. Details are in the document below. Please feel free to spread the word. Thank you!

📣 We are seeking a postdoctoral research fellow in ancient metagenomics. Join us at the Institute of Genomics, University of Tartu.

⏳ 01.04.2026–31.03.2028
📚 1.0 position
📍 Riia 23b, Tartu, Estonia

ℹ️ Project: lnkd.in/dQARuBxM
ℹ️ Application: lnkd.in/dFR8DQxH

⏰ Deadline 02.02.2026

Bias in genome-wide association test statistics due to omitted interactions Over the past two decades, genome-wide association studies (GWAS) enabled the discovery of thousands of variants associated with many complex human traits. However, conventional GWAS are still widely ...

Hello, just wanted to chime in that larger sample sizes can increase false positives under model misspecification. In a recent preprint, we showed that standard GWAS that omit epistasis yield more spurious associations as N grows, even when epistasis is minimal.
www.biorxiv.org/content/10.1...

Excited to announce a new 5-year project on developing generative models for complex trait genomics. I’m now looking for PhD candidates with CS, ML, stats or bioinformatics backgrounds. I’d very much appreciate it if you could spread the word. Contact: burak.yelmen@ut.ee

I think our results, in line with other recent findings, have important implications for GWAS methodology. Complex systems are rarely well captured by linear models; it’s reasonable to expect the same in complex trait genomics. Feedback on the manuscript is most welcome.

We show that even if the variance explained by epistasis is modest, it's possible to get false hits. The problem is also positively correlated with sample size, suggesting caution when interpreting significant associations especially for large-scale GWAS.

Bias in genome-wide association test statistics due to omitted interactions Over the past two decades, genome-wide association studies (GWAS) enabled the discovery of thousands of variants associated with many complex human traits. However, conventional GWAS are still widely ...

In our new study, we investigate how GWAS test statistics behave when a trait’s genetic architecture includes epistasis. We first model the bias mathematically and show via simulations how linear models can produce spurious significance (more in comments).
www.biorxiv.org/content/10.1...

There is a very interesting conference on ML for genomics in French Alps December 8th-12th this year. I will also talk about generative AI. Abstract submission for oral presentations is open till September 22nd.
legend2025.sciencesconf.org

𝐒𝐞𝐞𝐤𝐢𝐧𝐠 𝐚 𝐡𝐨𝐬𝐭 𝐢𝐧𝐬𝐭𝐢𝐭𝐮𝐭𝐢𝐨𝐧 𝐟𝐨𝐫 𝐲𝐨𝐮𝐫 𝐌𝐒𝐂𝐀 𝐏𝐨𝐬𝐭𝐝𝐨𝐜𝐭𝐨𝐫𝐚𝐥 𝐅𝐞𝐥𝐥𝐨𝐰𝐬𝐡𝐢𝐩?
Find your research group and join us at the University of Tartu!
Check our list of supervisors, application details, and everything else you need to know ut.ee/en/content/postdoctoral-research-university-tartu

Happy to announce an open call for a funded PhD position!

📊 Research project: Elucidating the fine-scale demographic history of the Eastern Baltic by incorporating haplotype-based analyses and modelling
📍 Institute of Genomics, University of Tartu
📩 Application period: 1st–15th May 2025

The Estonian Biobank’s journey from biobanking to personalized medicine - Nature Communications Large-scale biobanks have become a font of data that have led to discoveries across many fields of research. Here, the authors provide an overview of the Estonian Biobank, highlighting its value for r...

Excited to share our recently published overview of the @estbiobank.bsky.social 🧬
We highlight the unique features of the biobank and the conducted scientific research it has enabled, including the many recall studies that have been carried out over the years!
www.nature.com/articles/s41...