JACS Supplementary Cover - Week Commencing September 15

Another cover from our latest issue of #JACS: "Molecular Simulations with a Pretrained Neural Network and Universal Pairwise Force Fields"

Explore the research behind the art 🔗: buff.ly/J3R4JpL

#ChemSky

Grateful to everyone involved: @thorbenfrank.bsky.social, Sergio S. Dou, Almaz Khabibrakhmanov, Leonardo M. Sandonas, Oliver T. Unke, Stefan Chmiela, Klaus-Robert Müller, and Alexandre Tkatchenko.

This has been an enjoyable collaborative effort between @uni.lu, @tuberlin.bsky.social/@bifold.berlin, and Google DeepMind.

GitHub - general-molecular-simulations/so3lr: SO3krates and Universal Pairwise Force Field for Molecular Simulation SO3krates and Universal Pairwise Force Field for Molecular Simulation - general-molecular-simulations/so3lr

We hope this contribution will enable new simulations&insights and be of value to the community working on the next generation of general-purpose MLFFs.

The model, code and data are available here: github.com/general-mole...

To assess its capabilities and limitations, we benchmarked SO3LR on systems ranging from small molecules to large solvated biomolecules: protein, glycoprotein, and lipid bilayer. The model proved stable, reproduced local and global structural properties, and scaled to ~200k atoms on a single GPU.

It unites an equivariant neural network for semi-local effects with explicit physical potentials for short-range repulsion and long-range electrostatics/dispersion. The model was trained in <100 GPUh on a curated dataset of 4M molecular systems computed at the PBE0+MBD level of theory.

A persistent challenge in atomistic modelling is combining quantum-level accuracy with the efficiency needed for large, complex simulations. Our pretrained lightweight SO3LR MLFF is a step toward this goal.

Our recent work on SO3LR, a general-purpose machine learned force field for molecular simulations, has been published in @jacs.acspublications.org! 🌞 doi.org/10.1021/jacs...

Martini3-IDP: improved Martini 3 force field for disordered proteins - Nature Communications Here, the authors introduce Martini3-IDP, a refined model for disordered proteins that addresses prior over-compact structures. Validated across diverse systems, it captures IDP interactions and biomo...

Martini 3 - IDP is out! Improved parameters for disordered proteins, great work from Liguo Wang: www.nature.com/articles/s41...

Simulating full quantum mechanical ground- and excited state surfaces with deep quantum Monte Carlo by Zeno Schätzle, Bernat Szabo and Alice Cuzzocrea.

arxiv.org/abs/2503.19847

🧵⬇️

The QCML dataset, Quantum chemistry reference data from 33.5M DFT and 14.7B semi-empirical calculations - Scientific Data Scientific Data - The QCML dataset, Quantum chemistry reference data from 33.5M DFT and 14.7B semi-empirical calculations

This is a remarkable paper! A gigantic dataset of highly precise, highly accurate first-principles data. This builds on years of work on @fhi-aims.bsky.social - enabling dispersion-corrected hybrid DFT that covers a huge swath of chemical space. Congrats to the authors!

doi.org/10.1038/s415...

I rarely have been more excited about one of our papers. Come join the PET-MADness, to get better #ML potentials with less data 📈!

CHARMM-GUI EnzyDocker for Protein–Ligand Docking of Multiple Reactive States along a Reaction Coordinate in Enzymes Enzymes play crucial roles in all biological systems by catalyzing a myriad of chemical reactions. These reactions range from simple one-step processes to intricate multistep cascades. Predicting mech...

#compchem Good read: CHARMM-GUI EnzyDocker for Protein–Ligand Docking of Multiple Reactive States along a Reaction Coordinate in Enzymes #compchemsky #biosky pubs.acs.org/doi/10.1021/...

Analyzing Atomic Interactions in Molecules as Learned by Neural Networks While machine learning (ML) models have been able to achieve unprecedented accuracies across various prediction tasks in quantum chemistry, it is now apparent that accuracy on a test set alone is not a guarantee for robust chemical modeling such as stable molecular dynamics (MD). To go beyond accuracy, we use explainable artificial intelligence (XAI) techniques to develop a general analysis framework for atomic interactions and apply it to the SchNet and PaiNN neural network models. We compare these interactions with a set of fundamental chemical principles to understand how well the models have learned the underlying physicochemical concepts from the data. We focus on the strength of the interactions for different atomic species, how predictions for intensive and extensive quantum molecular properties are made, and analyze the decay and many-body nature of the interactions with interatomic distance. Models that deviate too far from known physical principles produce unstable MD trajectories, even when they have very high energy and force prediction accuracy. We also suggest further improvements to the ML architectures to better account for the polynomial decay of atomic interactions.

”Models that deviate too far from known physical principles produce unstable MD trajectories, even when they have very high energy and force prediction accuracy.”
doi.org/10.1021/acs....

Direct visualization of electric-field-stimulated ion conduction in a potassium channel Understanding protein function would be facilitated by direct, real-time observation of chemical kinetics in the atomic structure. The selectivity fil…

A fascinating paper elucidating ion conduction in a potassium channel experimentally

www.sciencedirect.com/science/arti...

We are seeking an outstanding #PhD candidate to join our team (www.duartegroupchem.org)! If you know potential candidates interested in combining #ML and #compchem reaction modelling for catalyst design, please share this opportunity with them!
🗓 Deadline: 31/01/25
📥 Application: shorturl.at/nHyAo

Excited to share our latest work on Euclidean fast attention, which enables learning global atomic representations at linear cost! 🔥

The representations describe the distance and orientation between atoms, crucial for modeling molecular systems

tinyurl.com/47xud8nr

#MachineLearning #AI4Science

SPINACH ON THE CEILING: A Theoretical Chemist's Return to Biology by Martin Karplus. #NMRchat #chemsky

www.annualreviews.org/content/jour...