@APoetsch

Computational biologist at TU Dresden. I am curious how mutations happen, why damage and mutagenesis differ a lot from place to place. @apoetsch.bsky.social

 Dresden, Germany
Joined June 2012
  • Tweets1.7K
  • Following615
  • Followers1K
106Media
Our 🧬 language model GROVER has now been published in @NatMachIntell. GROVER is a foundation model on the human genome to understand how such models learn grammatical structure in 🧬 and fine-tune it for more specific genomics questions. doi.org/10.1038/s42256-024-0… a 🧵 1/11

DNA language model GROVER learns sequence context in the human genome

Nature Machine Intelligence - Genomes can be modelled with language approaches by treating nucleotide bases A, C, G and T like text, but there is no natural concept of what the words would be and...

nature.com
8
29
1
163
13,981
Anna Poetsch retweeted
Anna Poetsch@APoetsch11 Apr 2025
Are you interested in combing machine learning and 🧬 language models with multi-omics data to study somatic genome evolution? We have several open postdoc positions and look forward to hearing from you! tu-dresden.de/cmcb/biotec/... 🧬🖥️

Biotechnologisches Zentrum (BIOTEC)

tu-dresden.de
1
7
9
1,468
Are you interested in combing machine learning and 🧬 language models with multi-omics data to study somatic genome evolution? We have several open postdoc positions and look forward to hearing from you! tu-dresden.de/cmcb/biotec/... 🧬🖥️

Biotechnologisches Zentrum (BIOTEC)

tu-dresden.de
1
7
9
1,468
There seems to be some overdue momentum to move to bluer skies. You can find me over there under @apoetsch.bsky.social 👋 (the algorithm didn't like this post before, so here it is again, but with a picture and some algorithm-candy: Immigrants rock! 💪🏻)
2
2
453
There seems to be some overdue momentum to move to bluer skies. You can find me over there under @apoetsch.bsky.social 👋
206
There’s a new Master in the house! Congratulations, Pablo! Excellent job 🥂
6
461
The book to read today (or yesterday)
3
11
2,221
Are you interested in how 🧬 works in the context of cancer and leukaemia development? We have several open postdoc positions. Interested? Please just get in touch. (I have not found the time to formally advertise them, yet)
2
4
1
9
2,122
more replies
One position is on 3D genome architecture in leukaemia using DNA language models.
1
234
One position is on genotype - phenotype relationships in glioma development using single-cell multiome data and machine learning.
235
PhD and Postdoc positions available! Interested in how our immune system evolves with a growing tumour, and how we can use that to fight cancer? Love data analysis? Applications now being accepted in the Machado Lab (machado-lab.org/contact) at the University of Cambridge.

Join Us — Machado Lab

machado-lab.org
7
154
5
454
74,732
If I were a postdoc again….
This quoted post is unavailable.
Try viewing anyway · View on X
1
1
249
I am happy to announce that as of today I will be Acting Chair for Genomics 🧬 @BIOTEC_TUD @tudresden_de. I feel very honoured for the opportunity! It is a temporary appointment that brings me one step further. Also, I guess, I sound more important now 🪑.
8
3
1
71
4,839
And I cannot even complain, because I did not manage to organise to vote. 🇦🇹, what are you doing? 😢
1
411
In our most recent paper published in @BMC_series #BMCBioinformatics, we looked further into what 🧬language models learn and proposed a fine-tuning task to evaluate foundation models doi.org/10.1186/s12859-024-0… a 🧵 1/8
4
3
11
1,045
What follows is a little experiment with the Twitter algorithm.
1
244
Immigrants rock! 💪🏻
1
231
Anna Poetsch retweeted
Anna Poetsch@APoetsch18 Sep 2024
In our most recent paper published in @BMC_series #BMCBioinformatics, we looked further into what 🧬language models learn and proposed a fine-tuning task to evaluate foundation models doi.org/10.1186/s12859-024-0… a 🧵 1/8
4
3
11
1,045
Big shoutout to the authors @SanabriaMeli and @JonasHirsch3! Well done! 🥳 (@SanabriaMeli also wrote this tweetorial, but the thanks is from me 😉)
1
148
In conclusion, next-kmer prediction requires learning of context beyond token identity. We can use it as a measure to evaluate 🧬 LLMs that will be used to discover new genome biology that goes beyond motifs and sequence content 7/8
1
104
By the way, we use the next-kmer task to choose the best vocabulary for GROVER doi.org/10.1038/s42256-024-0… 🧵 8/8
1
98
Load more