Michael Schubert
1.8K posts
Michael Schubert
@_ms03
👨🔬🧬 using 🧫 & 💻 to study how cells go rogue at @NKI_nl • Previously @EMBLEBI, PhD from @Cambridge_Uni • #cancer #aneuploidy #omics #rstats, dash of #hpc
Amsterdam, Nederland Bergabung Ağustos 2014
860 Mengikuti464 Pengikut
@deVisserKarin @Cancer_Cell @deWitLab @hannah_c_garner @m0rmart @NKI_nl @oncodeinstitute Congratulations! 🎉 This looks like a really interesting story
English
Very excited that our paper on how breast cancer impacts granulopoieses is @Cancer_Cell Fantastic teamwork with the lab of Elzo de Wit @deWitLab spearheaded by talented postdoc @hannah_c_garner (now jr groupleader at McGill) and PhD student @m0rmart @NKI_nl @oncodeinstitute
English
@EBItraining This is a great idea! Are the workshop resources somewhere accessible for non-participants?
(I can't find them on the Comptency Hub page)
English
It was great to be involved in this workshop focusing on exploring #AI competencies and training needs for research infrastructures.
Have you seen our competency hub that helps you explore career profiles and training resources? Find out more: competency.ebi.ac.uk
RItrainPlus@RItrainPlus
RItrainPlus has recently co-organised a workshop with @BiodevNetwork, @EBItraining & @sangerinstitute to explore the competencies necessary to support the use of AI in research environments. 🧪 Read more: ritrainplus.eu/2025/03/21/exp…
English
Michael Schubert me-retweet
Ralph Baric and Ian Lipkin in the NYT raising the alarm on a new Cell paper where Wuhan Institute of Virology scientists experiment on ANOTHER coronavirus that can infect humans under grossly inadequate biosafety precautions.
“could have significant risks for people worldwide.”
English
@paperpile Thanks! I would have never found this because it's not in the "Paperpile" top-level menu that I'm used to. I got this to work, but needed some attempts.
Sometimes the gear didn't show up, then the FAQ states a copy is made but the context menu that the current doc is edited
English
@_ms03 Hi Michael, thank you for reaching out! You can remove links from your Google Doc. See "What are citation links? How can I remove them?" in our FAQ: new.paperpile.com/FAQ-c9259d3444…
English
I love @paperpile, but it really needs an option to destructively remove all its links in a google doc #academicwriting
English
I just released version 0.9.7 of the #clustermq #rstats package: mschubert.github.io/clustermq/
_Run your code on HPC or via SSH via one line of code_
Please check it out; feedback is welcome!
English
New paper out! 🚨🥳
We thought that passenger gene amplifications can drive cancer vulnerabilities and show that RBM14 does so via cGAS-STING modulation 👇
@veronica_rendo @BenDavidLab @RameenBeroukhim @Foijer_lab
Veronica Rendo@veronica_rendo
So excited to see our paper describing the class of ARGOS genes in cancer out in @NatureComms! 🤩 nature.com/articles/s4146… This is the result of a giant effort led together with @_ms03, and executed under the mentorship of @BenDavidLab, @RameenBeroukhim, and @Foijer_lab. 🧵👇:
English
✨ Let me introduce you to my #PhDthesis ✨
I am glad to share that my #PhDthesis done at the @NKI_nl has been published and is freely available for anyone interested (see links below). Are you curious about the cover? Read the thread below
English
Michael Schubert me-retweet
📢 Applications for the NKI PhD program open from Jan 1 - Feb 28. Join our international research community in Amsterdam, working in cancer biology, AI research & life sciences with leading scientists and top facilities #PhD #CancerResearch More info: bit.ly/3O4bOjq
English
Michael Schubert me-retweet
I remain of the view that the disengagement of academics from this forum was a hugely self-defeating gesture. It scuttled what had been an unprecedented resource for academic communication and a valuable platform for public outreach.
Ethan Mollick@emollick
There are now papers showing that the migration of academic communities from Twitter to other social media sites mostly failed. But they never fully returned to Twitter either. I wonder if we will ever see so many different communities interacting in the same online space again.
English
Waiting for Godot ... to build in GitHub Actions. I'd really love to pay $$$ to get beefier build machines I don't have to maintain myself.
Windows, Linux, macOS, iOS, Android, web, release/debug, and a total of 8 variations for web.
Eeeks.
English
Michael Schubert me-retweet
"Quitting science", as if industrial research doesn't exist and/or there is shame in putting your *very* expensive degree to work in high level tech support, consultation, contract research, tutoring, entrepreneurship, etc, etc. Weird culty language to say leaving academia, tbh
nature@Nature
Nearly 50% of researchers quit science within a decade, huge study reveals go.nature.com/3Y9H9a9
English
Michael Schubert me-retweet
Our Review on “Aneuploidy as a driver of human cancer” is out in Nature Genetics. Check it out! 🎉
The Ben-David lab wishes everyone Shana Tova — may the new year be much better than the ending one, bringing peace and stability to all! 🙏
nature.com/articles/s4158…
English
So thrilled for the new chapter at VIB.AI and @KU_Leuven ! Beyond excited to contribute to their mission by diving deeper into the mysteries of the protein universe, working together to unlock new biology. Looking forward to what’s ahead in beautiful Leuven🚀🔬
VIB.AI@_VIB_AI
We're thrilled to announce @Joana_MPereira will join VIB.AI as group leader! Her pioneering research combines deep learning, network analysis, and bioinformatics to investigate protein diversity and evolution. 👉 lnkd.in/eemNRkm5 Welcome aboard, Joana!
English
Michael Schubert me-retweet
It's great to see my first paper of the PhD published in @NatureBiotech! go.nature.com/47ARizC. This describes our scRNA-seq method, scITD, for analyzing how the transcriptional states of multiple cell types are linked across patients. A few highlights from the paper:
English
@larsjuhljensen I understand all that, I just thought there was a second layer to "understanding the problem", which takes into account the time for heat to reach the center (your query is ambiguous in that regard)
In any case, thanks for engaging, I don't want to drag this out for too long 😅
English
@_ms03 It's very clear to me that almost all models look at my prompt and go "Aha, a math problem!" instead of recognizing it as a common sense prompt about cooking, i.e. boiling time does not depend on the amount of potatoes (but pointing out that size of them mattes would be a bonus).
English
The Qwen-2.5 14B #LLM is blowing my mind, especially with respect to actually following instructions. And I can run a well-performing Q4_K_M quantized version 100% on a GPU with 12GB of VRAM, giving me 32k context and 26 tokens/s output.
English
@larsjuhljensen Does the output change if you add "Assume same size of potatoes" or "Assume same number of potatoes"?
English
@_ms03 Haha, the exact phrasing of my prompt is pretty unambiguous, I think:
"If it takes 20 minutes to boil 200g of potatoes, how long will it take to boil 1kg?"
English
@larsjuhljensen Are we talking about 1 kg of 200g-sized potatos or one 1 kg sized potato here? 😆
English
The vast majority of models read it as a math question and assume direct proportionality between the amount of potatoes and the boiling time.
Curiously, the 14B version answered 40 minutes, so it did not fall into the usual math trap but still go tit wrong.
English