The course results were not only presented in our Institute Colloquium, but we worked on a detailed analysis of NMR conferences and published the results mr.copernicus.org/articles/6/2...

See also:
ista.ac.at/en/news/carb...

2) Trains do emit much less CO2 than planes per km and person, also when considering all the infrastructure. The savings can be an order of magnitude (and in most cases 3-4x, depending on the travel start/destination)
3) Per-person CO2 emissions to travel to a #EUROMAR #NMR conference is above 1 ton

In brief:
1) yes, carbon emissions due to conference travel are substantial (can be tens of tons of CO2 per year for some colleagues -- >10-fold more than the per-person "budget" to stay within a 2°C warming "goal", which in itself leads already to massive changes to our daily lives.

here is the correct link: ista.ac.at/en/job/tenur...

Preprint
www.biorxiv.org/content/10.1...

Data
doi.org/10.7910/DVN/...

Code
github.com/sai-advaith/...

8/8

All the credit goes to Alex Bronstein's team at @istaresearch.bsky.social. Amazing collaboration, largely driven by an amazing group: Advaith Maddipatla, Nadav Sellam Bojan, Meital Bojan, Volodymyr Masalitin, Sanketh Vedula, Ailie Marx and Alex. (thanx)

Preprint, data and code available👇
7/8

Guided AF3 may make ensemble-centric modeling routine in crystallography, NMR, and cryo-EM.
This includes proposing altlocs in X-ray, rapid ensemble interpretation in NMR, and injecting atomic detail into flexible cryo-EM regions.

6/n

2. X-ray/Cryo-EM: Electron density guidance can uncover previously unmodelled alternate conformations and restores structural accuracy in regions poorly predicted by unguided AF3, such as short peptides and flexible loops.
This methodology paves the way for experimentally aware predictive models
5/n

Guided AlphaFold3 resolves accurate ensembles from different experimental modalities:
1. Guiding with #NMR NOE distance or even order parameters (=atom-wise motional amplitudes) produces ensembles with less distance restraint violations and better replicates dynamic behavior.
4/n

This approach reframes protein structure determination as posterior inference and is fast! It takes data from #Xray, #cryoEM, #NMR (and more), and runs in several minutes, overcoming the severe computational bottleneck of traditional ensemble MD simulations. #StructuralBiology #AI #Protein
3/n

We present a new framework that treats AlphaFold3 as a powerful sequence-conditioned prior. By modifying the reverse diffusion steps of AF3, we inject a gradient-based guidance term derived from measured experimental data. This "steering" generates protein ensembles consistent with observations.
2/n

I am really enjoying this one every year. Together with a great group of students we analyse subjects related to sustainability. E.g. academic travel. We even got a nice little publication out of that work: mr.copernicus.org/preprints/mr...

This year: carbon footprint of computation and AI.

Check the last paper, equally pushed forward by Darja and @iomichiamofede.bsky.social in our lab @istaresearch.bsky.social
www.sciencedirect.com/science/arti....

The "L'Oreal For Women In Science" award will help Darja complete the final piece.

#NMR #Research #LOrealAward #FWF @fwf-at.bsky.social

The specific labeling allows seeing very important functional arginines even in >100 kDa large proteins. In MAS #NMR, we probe Arg dynamics without size limits.
Special shoutout to Fede @iomichiamofede.bsky.social , a great scientist and our master of enzymes and NMR in solids and solution.

3/n

Darja combines organic-chemistry synthesis (with our great collaborator Roman Lichtenecker) of specifically isotope-labeled amino acids with protein NMR in solution and solids.
👉 Synthesis and proof of principle is here: chemistry-europe.onlinelibrary.wiley.com/doi/full/10....

2/n