Posts on X

Interesting paper from @TheBoutrosLab on a spatial expression atlas (in drosophila wing disc) using scRNA-seq and colocalizing with known expression patterns. Looks like extension of spatial anchoring methods by @MarioniLab & @satijalab, missed comparison. rdcu.be/bMOml

Great short and accessible read for improving your #MachineLearning model.

@khanaziz84

Three pitfalls to avoid in #machinelearning nature.com/articles/d4158…

Quoted post

Here the open access readcube link to the paper rdcu.be/bMhnD

@fabian_theis

Happy that our tool scGen just got published in @naturemethods. We show that generative models predict single-cell perturbation effects out-of-sample - great work led by @MohammadLotfol1 and @falexwolf! Try it at github.com/theislab/scgen, read at nature.com/articles/s4159…

Quoted post

Happy that our tool scGen just got published in @naturemethods. We show that generative models predict single-cell perturbation effects out-of-sample - great work led by @MohammadLotfol1 and @falexwolf! Try it at github.com/theislab/scgen, read at nature.com/articles/s4159…

Hard at work @HelmholtzMucEn summer party with @EleZeggini 😅

Congrats @TU_Muenchen for getting @dfg_public‘s ‚Excellence University‘ label again!

GIF, plays on X

@Mkg_Lehmann

Skydiving into epithelial differentiation dynamics. Happy to be part of this awesome project from @SchillerLab +@fabian_theis lab. Sequencing of more than 30000 epithelial cells. This will be a great resource to understand and #cureIPF @m_konigshoff biorxiv.org/content/10.110…

Quoted post

Dr. Büttner @marenbuettner

@CompHealthMuc

Congratulations Dr. Maren Büttner @marenbuettner @fabian_theis

Quoted post

Facebook's Poker AI outperforms humans in multiplayer poker. This kind-of kills online poker.. science.sciencemag.org/content/early/…

Great lecture for high-school students by @comedyhacker Tobias Schrödel, talking amongst others about Grace Hopper, ‘Grandma of Cobol’, who found the first actual computer bug in the late 40s.

‘PhD recipients should feel confident in their career potential, especially if they are willing to look beyond universities.’

@FreethePhD

Science PhDs lead to enjoyable jobs freethephd.com nature.com/articles/d4158…

Quoted post

Great to be able to systematically measure RNA velocity in single cells (‘new to total RNA velocity’). Really exciting new protocol, congrats Florian, @AE_Saliba and @DolkenL, and thank you for this collaboration

@AE_Saliba

Our scSLAM-seq paper is out in Nature @Nature! It is an amazing collaboration with Lars Doelken @DolkenL and Florian Erhard @Uni_WUE with the help of Fabian Theis @fabian_theis. Thousands thanks guys! nature.com/articles/s4158… #hiri #singlecell

Quoted post

At a recent lecture, when introducing k-means, I suggested a hands-on ‘algorithm dance’: essentially we had 2 students as iteratively updated centroids - the others were samples, who pointed to their closest centroid. Any other suggestions for interactive #machinelearningdances?

"We predict that researchers who are capable of applying machine learning to complex biological data will be increasingly in demand in the future." Amen to that! Short editorial on #machinelearning in biology @GenomeBiology, see rdcu.be/bJrGC

Beautiful summer party @ICBmunich ❄️🐝 - so grateful for the honor of working with these clever, cool, sometimes nerdy and always nice people!

This agrees with what we found, too: biorxiv.org/content/10.110… However we did validation on independent data sets (different labs) instead of internal cross validation, which may reflect application use cases better. t.co/3n6SzeeHHA

Quoted post

Good Q on experimental design: even in the limit of average 1 read/cell there’s s.t. to be learned namely univariate distributions of each gene, which you could compare differentially across conditions. No multivariate info (cell types...) though. Anyone knows any papers on this? t.co/SlhF1NMvfp

Quoted post

Just saw this from Peter: 10^13 cells in the human body - don’t just average them. Love it!

@KharchenkoLab

A quant in search of a worthy problem? Consider our postdoc vacancies! pklab.med.harvard.edu

Quoted post

Thanks for your super interesting talk on ‘simulation-based inference’ (Aka ABC 🤔), great discussion plus some sports!

@jakhmack

@ICBmunich @TU_Muenchen I had a fantastic day yesterday and thoroughly enjoyed the discussions with @fabian_theis and all the other @ICBmunich members about their exciting science -- many thanks for the invitation!

Quoted post

Really looking forward to this @humancellatlas collaboration with Regev and Teichmann labs, @dana_peer, @SchillerLab, @KaminskiMed, @Eickelberg_MD, Barbry lab and many others. Thx @Misharin_lab for coordinating all of this, and of course @cziscience for funding!

@SchillerLab

Happy to receive funding from the Chan Zuckerberg Initiative together with @fabian_theis. We participate in the seed network building a Human Lung Cell Atlas 1.0 👍@humancellatlas @HelmholtzMucEn @cziscience @Misharin_lab @singlecellomics bit.ly/2WYc9XC

Quoted post

Great - now please create the corresponding AI to write my grant proposals!

@welcomeai

Google has created a maths AI that has already proved 1200 theorems. Stay up to date on AI in your industry at hubs.ly/H0hFsQf0 #artificialintelligence #google #maths #ai #industry #strategy #algebra #research #business #intelligence #marketing #entrepreneur #data #ceo

Quoted post

Very useful, thx! Also the recommendations about choice of particular performance metric.

@markrobinsonca

Our thoughts on benchmarking computational methods .. bit.ly/2IuFSUo

Quoted post

Congrats, Caro!

@JanHasenauer

Congrats Dr. @carolin_loos! Great PhD defence.

Quoted post

Congrats, Michael - great effort, those community challenges are very useful and results interesting.

@MendenMichael

Finally released, thanks all for the hard work :-) nature.com/articles/s4146…

Quoted post

Thanks for this great collaboration! Interesting to see how these scRNA-seq atlas projects now are increasingly able to incorporate perturbations or even disease settings.

@MC_Nawijn

Happy to see our @HCA paper on Lung and asthma cells types and states finally appear: nature.com/articles/s4159… Great collaboration with Sarah Teichmann's team @sangerinstitute, Herbert Schiller & Fabian Theis @HelmholtzMucEn and @GSK Thanks @Longfonds for supporting our research!

Quoted post

Hey, really nice to see that some people are still looking into this somewhat older book! Now let’s convince him to also check out spatial transcriptomics ;)

@DenisSchapiro

Our rotation student @HiTSatHarvard from @hseas showed me the book which convinced him to check out biomedical imaging. Thank you @fabian_theis

Quoted post

Our tutorial on the lab’s current best practices for single‐cell RNA‐seq analysis @MolSystBiol is out. Great effort by @MDLuecken! Read at bit.ly/2L2r018, and please also check out the joint python&R analysis workbook github.com/theislab/singl…

Deep learning does not outperform classical machine learning for cell-type annotation. Recent short finding led by Niklas and @marenbuettner, where we wondered why only unsupervised learning is used in single-cell genomics. Need more complex annotations! biorxiv.org/content/10.110…

a truly useful and very promising resource for genomic deep learning - congrats!

@gagneurlab

The @KipoiZoo paper is out now, with substantial progress to ease model submission since the biorxiv preprint. Try it and share your models. Fantastic collaboration indeed -- open to more contributors! kipoi.org

Quoted post

Feedback by our advisor @M_Feindt @BYAnalytics at @helmholtz_en data science workshop - continue to invest in #AI, but remember it‘s more than just #DeepLearning. @helmholtz_ai