Sebastian Schönherr

This is what I need today.

Last November, we migrated the Michigan Imputation Server (imputationserver.sph.umich.edu) to a new software stack, including Nextflow. After some turbulence, we are more and more returning to normal operations, processing around 800.000 full human genomes per month.

Today @michaelmoor.bsky.social took a train from Zurich to Innsbruck to kick-off our new Faculty of "AI and Scientific Computing"! 🎉 He talked about LLMs and Medical AI Agents. Exciting science and great discussions! Thanks! More info about our faculty: aiscm.i-med.ac.at #ai #scientificcomputing

Silvia is excited, we are so excited. Follow Silvia for great science and great fun.

www.youtube.com/watch?v=8iwB...

🚀 Ready to kick off our Nextflow in Genomics workshop tomorrow at @i-med.ac.at!
🎉 Together with @lukfor.bsky.social & @silvidm.bsky.social we'll work on improving imputation server workflows.

Big thanks to @nextflow.io, @evanfloden.bsky.social, and @ewels.bsky.social for the support.

If you didn't hear back from me today, it's because I had to solve a 4th grade math test first.

Almost forgot: Shout out to @nextflow.io for sending us stickers to get a bit of conference vibe here in innsbruck 🎉

Nextflow In Genomics 2025 - Genepi Innsbruck Institute of Genetic Epidemiology, Medical University of Innsbruck

Details about the workshop can be found here: genepi.i-med.ac.at/workshops/ne... Please feel free to reply directly to this thread for feature requests or via Github: github.com/genepi/imput...
2/2

Nextflow In Genomics 2025 - Genepi Innsbruck Institute of Genetic Epidemiology, Medical University of Innsbruck

Are you working with one of the genotype imputation servers? Then this is your chance to see new features pretty soon on the server: @lukfor.bsky.social ky.social, @silvidm.bsky.social, and I are hosting our 2nd Nextflow in Genomics Hackathon to boost imputation servers here @i-med.ac.at
1/2

Publishing software is often easier than maintaining it - but we’re committed to keeping ours up to date! 🛠️ We've enhanced the documentation for our mtDNA variant calling pipeline, with detailed guides for both graphical and command-line use with Nextflow:
mitoverse.readthedocs.io/mtdna-server...

Btw, you can now set the mean minimum coverage when submitting a job, making it easier to work with low-coverage data. Let us know which features you’d like to see next. For graphical job execution click here: mitoverse.i-med.ac.at

Publishing software is often easier than maintaining it - but we’re committed to keeping ours up to date! 🛠️ We've enhanced the documentation for our mtDNA variant calling pipeline, with detailed guides for both graphical and command-line use with Nextflow:
mitoverse.readthedocs.io/mtdna-server...

endless slopes

And I almost forgot the most frightening moment in this journey: shutting down the old server after 9 years. #byebyehadoop

Our framework AND pipeline are now used by all the major imputation servers - Michigan, Topmed, Munich - which makes us quite proud. It’s also great to see @stephenturner.us covering us in his recent blog: tinyurl.com/5tp9tz7e.
We would love to hear from more imputation servers around the world. 🌏

This summer, @lukfor.bsky.social and I migrated the Michigan Imputation Server to Nextflow. 2M genomes later, we can say it was a big success.

As a Christmas gift, we’re excited to show you how to set up your own local server in under 10 minutes.

##freeimputation

genepi.github.io/michigan-imp...

Last week was fun! We had the pleasure of welcoming a MINT high school to our institute to talk about A.I 🤖. We also dived into Genetics (by @stncsn.bsky.social), Medicine, and discussed their project on AI-powered shark detection. 🧬🩺🦈 #STEM #MINT #Education #AI #Genetics @meduniibk.bsky.social

Today @silvidm.bsky.social delivered an impressive talk on her latest research on repetitive regions in the human genome - could they hold the key to the missing heritability? Huge thanks to the Biocenter @meduniibk.bsky.social for hosting this. #ukb #nextflow #lpa #vntrs

We believe this work will help others analyze VNTRs from short-read sequencing data and understand their impact on numerous human phenotypes, potentially explaining parts of the missing heritability. end.

For the first time, we were able to analyze the LPA VNTR in >199,000 samples from the UKB. This approach successfully revealed new, strong Lp(a)-lowering effects for KIV-2 variants, with a protective effect against coronary artery disease, and also validated previous findings. 5/n

And yes, we can! We developed a computational approach to resolve intra-repeat variation in the KIV-2 VNTR region using widely available short-read sequencing data, and we also applied it to 5 other medically relevant VNTRs 4/n

We specifically focused on the "dark" region of the cardiovascular disease gene LPA – the so-called KIV-2 VNTR. We wanted to determine if we could decipher this region using short-read sequencing data 3/n

Resolving intra-repeat variation in medically relevant VNTRs from short-read sequencing data using the cardiovascular risk gene LPA as a model - Genome Biology Background Variable number tandem repeats (VNTRs) are highly polymorphic DNA regions harboring many potentially disease-causing variants. However, VNTRs often appear unresolved (“dark”) in variation d...

In July this year, we (Silvia Di Maio, Peter Zöscher and @stncsn.bsky.social et al.) published a Genome Biology paper to decipher repeats in the human genome, so called VNTRs genomebiology.biomedcentral.com/articles/10.... 2/n

(c) Medical University of Innsbruck, Cristof Simon

A typical shot from my life as a researcher at @meduniibk.bsky.social. Standing in the HPC center with Silvia Di Maio and looking at dark genome regions

What's the story behind? 1/n

🌍 The imputation servers running on our software stack:

🛰️ Our Main Instance: imputationserver.sph.umich.edu
🛰️ TOPMed: imputation.biodatacatalyst.nhlbi.nih.gov
🛰️ MUC: imputationserver.helmholtz-munich.de

💡 Are there other imputation servers flying around?

Cloudgene 3: Transforming Nextflow Pipelines into Powerful Web Services Nextflow has emerged as a flexible and scalable workflow system for building computational pipelines in life sciences. Simultaneously, Software-as-a-Service platforms enable researchers to concentrate...

Check out @lukfor.bsky.social and my latest work to run Nextflow graphically. Cloudgene3 powers major genotype imputation servers (Michigan, TOPMed, Helmholtz Munich—you name it).

It’s like a referee in football—when no one is complaining, you know it’s doing an excellent job! ⚽ 🧬
shorturl.at/pdeGk