Philippe Schwaller

Bulk Measurement of Membrane Permeability for Random Cyclic Peptides in Living Cells to Guide Drug Development Some cyclic peptides can cross membranes, which is highly attractive for drug development, but it remains difficult to accurately predict membrane permeability or to test it experimentally for large ...

Very pleased to share the 'last' first-author paper from my postdoc with Prof. Christian Heinis at EPFL.
Now out in its final form in @angewandtechemie.bsky.social as a 'Very Important Paper' 🔥 (1/🧵) #chemsky
onlinelibrary.wiley.com/doi/10.1002/...

Finally, we teamed up with @pschwllr.bsky.social @rebeccaneeser.bsky.social who prepared a random forest model that can readily applied to estimate the membrane permeability of designed cyclic peptides.
Model available on GitHub: github.com/schwallergro...
(8/🧵)

Burgenstock #bc_scs25 @swisschemistry.bsky.social Editorial Board meeting R. Buller @albrecht-lab.bsky.social @olallalab.bsky.social D. Katayev
@harigroupiisc.bsky.social @remglasgow.bsky.social
@evaheviagroup.bsky.social J. Lacour
and

Yay for @pschwllr.bsky.social and @mlederbauer.bsky.social (and all your co-authors who aren't on BlueSky yet) 🥳

This #dataset is a prime example of #GoodData, and it ties nicely with what @clarakirkvold.bsky.social and @grynova.bsky.social were talking about a few weeks ago in their #JournalClub

Congratulation @m-iwan.bsky.social aichemist.eu/fellows with the Best Oral presentation award during 21st International Workshop on Quantitative Structure-Activity Relationships in Environmental and Health Sciences (QSAR2025) qsar2025.marionegri.it Looking forward towards your publication Mateusz!

Last week to apply to this position in my group.

EPFL and ETHZ in Switzerland are also top universities.

No way to investigate that as ChatGPT is closed source.

In general, LLMs hallucinate by mixing learned patterns with gaps in knowledge, generating plausible-sounding but fake details confidently.

Training large language models for chemistry is bottlenecked by one critical problem: there is no unified dataset that connects all chemical domains.

Friends from the #AcademicSky, one of our students is interested in studying the acceptance of LLM use in social media. Can you spare a couple of minutes to answer their survey?

survey.unifr.ch/index.php/38...

A variety of responses would mean the world to us!

@unifr.bsky.social

We demonstrate granular and steerable synthesizability control in molecular generation, allowing to specify reaction and building block constraints while optimizing for molecules with desired properties.

Super proud of Jeff Guo and Víctor Sabanza Gil leading this exciting study! #chemsky

Generate property-optimized small molecules with 𝘴𝘵𝘦𝘦𝘳𝘢𝘣𝘭𝘦 𝘢𝘯𝘥 𝘨𝘳𝘢𝘯𝘶𝘭𝘢𝘳 synthesizability control - allowing complete user-flexibility to impose various reaction constraints!

Pre-print: arxiv.org/abs/2505.08774
Code: github.com/schwallergro...

(1/4)

Just wow @evaheviagroup.bsky.social at #BC_25

JSP program at the Bürgenstock conference 2025 promotes Scientific Leaders of the next Generation from both academia and industry! Class 2025 JSP together with the Bürgenstock president José Luis Mascareñas. #SCS_BC25 #JSPFellow2025 #scs

The final day of the AIChemist-CECAM School @cecamevents.bsky.social has started - time flies when you're having fun! @pschwllr.bsky.social recaps the work done over recent years towards developing complementary AI systems that bridge the gap between computational design and experimental reality.

The school "AIChemist" has just started in CECAM-HQ!
After a short welcome by the organisers, Jean-Louis Reymond will start today's program with a talk on "Expanding chemical reaction space for synthesizing the GDB".
➡️ www.cecam.org/workshop-det...

What a fantastic line-up!

But wait, there's more! 💫 Check out the full speaker list at www.cecam.org/workshop-det....

Registration is open until the 28th of March 📅 Don't miss out!

The AiChemist CECAM School starts with lecture of Prof. J.L. Reymond www.dcbp.unibe.ch/ueber_uns/personen/prof_dr_reymond_jean_louis/ from the Uni Bern with description of challenges to synthesise theoretically predicted compounds from GDB database. See whole program: www.cecam.org/workshop-det...

Flow-Based Fragment Identification via Contrastive Learning of... Fragment-based drug design is a promising strategy leveraging the binding of individual fragments, potentially yielding ligands with multiple key interactions, surpassing the efficiency of full...

Hello from Singapore 🇸🇬! Thrilled to be at #ICLR2025 presenting our work on fragment-based drug discovery 🧩. We go beyond virtual screening with a generative, structure-aware approach.

📃 openreview.net/forum?id=bZW...

🔗 github.com/rneeser/Late...

A thread 🧵👇

In this case, we didn't run DFT calculations but built the GMAE benchmarks from existing datasets that typically report multiple local minima per adsorbate.

Check out Rxn-INSIGHT, an open-source tool that can help chemists to rapidly and accurately classify reactions, name compounds, and recommend conditions for novel syntheses.

Docs: rxn-insight.readthedocs.io/en/latest/
Paper: doi.org/10.1186/s133...

What began as Xu's undergraduate thesis in Spring 2023 evolved into an impressive publication after multiple improvement rounds. Not every undergrad project leads to such a quality paper!

A multi-modal transformer for predicting global minimum adsorption energy - Nature Communications The fast evaluation of global minimum adsorption energy (GMAE) is crucial for catalyst screening. Here, authors designed a multi-modal transformer called AdsMT to rapidly predict the GMAE without site...

Thrilled to announce our new paper on predicting global minimum adsorption energy (GMAE) for computational catalyst discovery! We developed a multi-modal transformer approach, including adsorption site identification via cross-attention. Huge congrats to Junwu and Xu! www.nature.com/articles/s41...

Raphael Bigler F. Hoffmann La Roche and Philippe Schwaller @pschwllr.bsky.social present their collaboration ‘Combining High-Throughput Experimentation (HTE) and Data Science to Optimize Chemical Processes’ @swisschemistry.bsky.social #fs25 @hes_so

YouTube video by Andrew S. Rosen How to Easily Make a Python Package (batteries included)

How to Easily Make a Python Package (batteries included) with Professor Andrew Rosen @andrewrosen.bsky.social
www.youtube.com/watch?v=th2C...

NeurIPS participation in Europe We seek to understand if there is interest in being able to attend NeurIPS in Europe, i.e. without travelling to San Diego, US. In the following, assume that it is possible to present accepted papers ...

Would you present your next NeurIPS paper in Europe instead of traveling to San Diego (US) if this was an option? Søren Hauberg (DTU) and I would love to hear the answer through this poll: (1/6)

Check out our paper published in @chemicalscience.rsc.org - pubs.rsc.org/en/content/a...

(1) *General-purpose* generative model + retrosynthesis model = design molecules with optimized properties with an *explicit* predicted synthesis pathway

(1/2)

The thematic issue “Adaptive experimentation and optimization in organic chemistry” edited by Philippe Schwaller @pschwllr.bsky.social and Artur M. Schweidtmann is complete ✅.

Download the issue PDF for free with just one click: ➡️ https://www.beilstein-journals.org/bjoc/series/131?B=y

💎🔓 #BJOC

New preprint from the Schwaller group!

We investigate the chemical reasoning of LLMs on synthesis routes (up to 26 steps) and reaction mechanisms. Claude 3.7 performs impressively well.

- LLMs for direct outcome prediction ❌
- Search algorithms + LLMs 🤩

Congrats to the team! 🙌

#chemsky ⬇️

LLMs are pretty bad at writing molecules, but quite good at analyzing mols and reactions!
In our new work we use LLMs+search in chemical tasks, unlocking steerable synth. planning and mechanism prediction 🌟

Chehck out the paper 👉 arxiv.org/abs/2503.08537

1/8