marwinsegler.bsky.social - Profile | ThreadSky | a Reddit-style client for Bluesky

Principal researchers inside and outside academia: do you know groups self-organizing to protect academic science in the US? How can we connect and mobilize? Could you share your thoughts? 🧬🔬🔭🧪🧠🤖

submitted 4 days ago • 5 comments

Wish more people were saying this

submitted 3 days ago • 0 comments

Geht’s raus und geht’s wählen! ⚽️🗳️

submitted 3 days ago • 0 comments

Prediction: In the not so distant future when the models John describes will be out, people will talk about the „Achira-moment“ for CADD!

submitted 5 days ago • 0 comments

We start our first post on Bluesky with a firework! Very proud of a brilliant team to publish our work on C(sp3)-atom transfer @science.org. www.science.org/doi/10.1126/... It has been a very exciting journey. Thanks @erc.europa.eu

submitted 6 days ago • 6 comments

We made a new, reproducible, fair, (and much harder) motif scaffolding benchmark! Paper: arxiv.org/abs/2502.12479 Code: github.com/blt2114/Moti...

submitted 7 days ago • 2 comments

The BioEmu-1 model and inference code are now public under MIT license!!! Please go ahead, play with it and let us know if there are issues. github.com/microsoft/bi...

submitted 7 days ago • 3 comments

The future of CADD is coming!

submitted 7 days ago • 0 comments

Cheminformatics job @chemjobber.bsky.social roche.wd3.myworkdayjobs.com/roche-ext/jo... #compchem #cheminfo #chemsky

submitted 7 days ago • 0 comments

Professors please pull up a chair. I’ve been talking to my classes about what has been going on. I explained indirect costs to them. I talked to them about what a probationary employee is in the government. At the end of class they asked if we could talk about it more. (1/)

submitted 8 days ago • 91 comments

1/3 In a new preprintlet, we show that representing mass spectra as path graphs works well. In a regression task predicting QED, mass spectra line graphs + GAT performed 24% better (MAE) than binning + MLP. With @jjjvanderhooft.bsky.social and Niek de Jonge. arxiv.org/abs/2502.10851

submitted 8 days ago • 2 comments

We're looking for Research Engineers to work at the intersection of LLMs and impactful scientific problems - see boards.greenhouse.io/deepmind/job... for more details. Applications close next Tuesday 18th.

submitted 15 days ago • 1 comment

Diffusing crystals looks like dancing!

submitted 15 days ago • 0 comments

The transition-metal catalyst reduced the activation barrier to convert the cis into the trans isomer.

submitted 15 days ago • 0 comments

It’s been a tough few weeks. My 10yo daughter was diagnosed with a very rare, aggressive cancer called interdigitating dendritic cell sarcoma (IDCS). I’m reaching out to identify clinicians/patients who have encountered pediatric IDCS or other (non-LCH) dendritic or histiocytic sarcomas cases.

submitted 18 days ago • 86 comments

Fantastic to have the output of this unique conference as a real book!

submitted 21 days ago • 0 comments

Only five more days to apply!

submitted 31 days ago • 0 comments

Open roles at MSR AI for Science: ML Research & ML Engineer If you passionate about Deep Learning and want to invent the future of scientific discovery jointly with an amazing team, please apply! jobs.careers.microsoft.com/global/en/jo... jobs.careers.microsoft.com/global/en/jo...

submitted 34 days ago • 0 comments

I have a new blog devoted to medicinal chemistry in which I will mainly focus on classic papers. My first installment describes the design and discovery of the amazing atorvastatin: a triumph of simple modeling, dogged synthesis and leadership decisions medchemash.substack.com/p/classics-i...

submitted 40 days ago • 2 comments

#compchem

submitted 41 days ago • 0 comments

📢 Paper + code release 📃💻 After 2 years of work, I'm excited to announce our newest paper, MatterGen, has been published in Nature! www.nature.com/articles/s41... We are also releasing all the training data, model weights, model code, and evaluation code on GitHub! github.com/microsoft/ma...

submitted 41 days ago • 3 comments

#compchem

submitted 43 days ago • 0 comments

If you use generative #DeepLearning for molecule design, check out our latest work, where we perform a large scale analysis (~1 B designs!) and find ‘traps’, ‘treasures’ and ‘ways out’ in the jungle of generative drug discovery. 🌴 🐒 Paper: arxiv.org/abs/2501.05457 Code: github.com/molML/jungle...

submitted 43 days ago • 0 comments

Reaction outcome prediction with AI models could be a powerful capability in workflows, including the exciting prospect of discovering new reactions! But how well do they work on more realistic data? Check the preprint to find out! arxiv.org/abs/2501.06669 and see Connor’s long LinkedIn post:

submitted 43 days ago • 2 comments

Awesome paper! BioEmu from MSR: Scalable emulation of protein equilibrium ensembles with generative deep learning www.biorxiv.org/content/10.1... Looking forward to hearing more explanainers in tomorrow's reading group session on zoom at 9am PT / 12pm ET / 6pm CET: portal.valencelabs.com/logg

submitted 45 days ago • 1 comment

Thanks to those who’ve reached out the last few days. I am grateful that my family and home are safe. Many of my friends and colleagues have lost homes. If you want to help Caltech/JPL faculty/students/staff, a relief fund has been established: giving.caltech.edu/areas-to-sup...

submitted 46 days ago • 2 comments

Interested in building the future of open source ligand- and structure-based ML models for ADMET prediction? The ARPA-H funded OMSF (@omsf.bsky.social) OpenADMET project is hiring multiple positions to build ML models of ADMET properties and drive informative data collection! openadmet.org/jobs/

submitted 49 days ago • 0 comments

Meet autoplex – our approach to automated ML potential fitting, built jointly with @molecularxtal.bsky.social & team in Berlin! In this preprint, we focus on exploring structures and training potential models "from scratch" with the help of automated workflows: arxiv.org/abs/2412.16736

submitted 50 days ago • 1 comment

Molecular simulations or AI models produce ever larger and more complex datasets. Comparing different conditions to find meaningful patterns is a bottleneck for many projects. We address this with PENSA, a flexible open-source analysis library. Now out in JCP: doi.org/10.1063/5.02... 🧪 #CompChem

submitted 50 days ago • 1 comment

In case your new year's resolution is to do a PhD—there's still time until the 6th to apply 😉

submitted 57 days ago • 0 comments