Multi-Scale Protein Structure Modelling with Geometric Graph U-Nets 1. A new class of models, Geometric Graph U-Nets, has been introduced to capture the hierarchical nature of protein structures by learning multi-scale representations through recursive coarsening and refining of protein graphs. This approach mirrors the biological hierarchy of proteins, from local motifs to global domains. 2. The study demonstrates that Geometric Graph U-Nets are theoretically more expressive than standard Geometric GNNs, as proven by extending the Geometric Weisfeiler-Leman (GWL) test. This hierarchical design allows the model to capture both local interactions and long-range effects crucial for protein function. 3. Empirically, the model shows substantial improvements in protein fold classification tasks, outperforming both invariant and equivariant baselines. This highlights the model's ability to learn global structural patterns that define protein folds, integrating both global context and local geometric details. 4. The paper introduces two novel pooling layers—Point Pooling and Sparse Pooling—which are designed to be geometrically meaningful, computationally efficient, and compatible with both invariant and equivariant Geometric GNN layers. These layers enable the construction of deep, multi-level hierarchies for protein graphs. 5. The authors provide a rigorous theoretical foundation for the expressivity of geometric pooling layers, showing that pooling can not only preserve but also enhance the model's ability to distinguish non-isomorphic geometric graphs. This is empirically validated through experiments on synthetic k-chain graphs. 6. The research opens new avenues for designing geometric deep learning architectures that can effectively model the multi-scale structure of proteins, with potential applications in structural biology and drug design. Future work includes exploring the model's performance on a broader range of protein-related tasks and extending it to dynamic biomolecular systems. 📜Paper: arxiv.org/abs/2512.06752v1 #ProteinStructure #GeometricDeepLearning #GraphNeuralNetworks #MultiScaleModelling #StructuralBiology

🌟 DAIMON at @MWMaterialsWeek 2024 in Cyprus! 🌊 We had the opportunity to present our multiscale modelling approach for SSbD materials and applications in Limassol! Thanks to the organizers and all participants! #AdvancedMaterials #MultiscaleModelling #SSbD #MaterialsModelling

Molecular dynamics and multiscale modelling for sustainability @Daimoners Full time researcher in #moleculardynamics and #multiscalemodelling for sustainable materials @CNR-ISMN Bologna, Italy #Gromacs #LAMMPS jobrxiv.org/job/daimon-team-… #gromacs #lammps... jobrxiv.org/job/daimon-team-…

Had some interesting discussions regarding long-range interactions in polymer simulations among various other topics #PULSCappricio, it was a pleasure having @mxvassaux over at PULS group @UniFAU . Looking forward to more interesting exchanges. #multiscalemodelling

Alla scoperta di #ICSC💻 L'ambizione dello #Spoke6 #MultiscaleModelling and #EngineeringApplications è riuscire a comprendere, grazie all'#HPC e ai #BigData, quello che accade in fenomeni fisici caratterizzati da scale diverse [1/2] @SapienzaRoma @Unipisa @mur_gov_

👍 Great program! Contratulations! We are proud having Maxim Igaev @maxotubule from our group @mpi_nat contributing with at talk on day one #MultiscaleModelling to the GGMM2023 coming up soon ggmm2023.sciencesconf.org/re… 🥳

A computational model of #Streptomyces reveals that if antibiotic production trades off with replication, genome architecture can evolve to support a mutation-driven division of labor ➡️ bit.ly/3XZv7gT @ESColizzi @RenskeVroomans #evolution #multiscalemodelling

🗓️Today at 3 pm join the #NanoColloquia2023 by Alessandro Mossa about “Multiscale modelling applied to the development of therapeutics for COVID-19” To watch the streaming 👇 tinyurl.com/AlessandroMossa More info nano.cnr.it/seminar-cnr/mult… @CNRsocial_ #MultiscaleModelling

#ICSCsupercomputing aggregherà le eccellenze scientifiche italiane in 10 diversi ambiti: #HPC e #BigData, ricerca fondamentale e space economy, astrofisica, clima, ambiente, #MultiscaleModelling, scienze dei materiali, medicina computazionale, #SmartCites e #QuantumComputing.

Registration is now open for the 5th @ccpbiosim / @CCP5UK Conference on #MultiscaleModelling of Condensed Phase and Biological Systems... 3-5 April 2023, Manchester, UK... ccpbiosim.ac.uk/multiscale20… ⬇️⬇️⬇️

👩‍🔬Want to learn how to use a range of biological #modeling and #simulation software at a cellular level? @PerMedCoE #SummerSchool is for you. 🗓️Save it! June 25-30, 2023: bit.ly/3BxE3jF #CellSimulations #MultiscaleModelling #HPC #DataAnalysis #ML #HealthTech #biotech

Kick off meeting of Spoke 7 #Materials and #molecular sciences of the new National Center for #HPC, #BigData and #QuantumComputing. @PoliTOnews and @musychen will be involved in several activities related to #multiscalemodelling! Good luck to all of us!

#ECCB2022 workshop NTB-W03 | Simulating cellular behaviours: advancing HPC-enabled Computational Biology 💻 Online for registered participants about to start eccb2022.org/ntb-w03 #CellSimulations #MetabolicModelling #MultiscaleModelling #HPC #DataAnalysis #ML

🚨New member alert🚨 Meet Yilin! His research is in the field of #multiscalemodelling of #thrombolytictherapies for the treatment of ischemic stroke ⚙🩸 Welcome to the team Yilin 😁

Happy to be part of this nice work and this amazing nanosafety community! #nanoparticle #proteincorona #multiscalemodelling #metadata #metadatacompleteness

Exciting weeks ahead! :) #MBNExplorer #MultiscaleModelling

For our next #InsigneoSeminar, Carlos Borau from the University of Zaragoza is visiting the institute to give an in-person talk on 'Microfluidics and multiscale modeling in mechanobiology' on 17/06/22 sheffield.ac.uk/insigneo/ove… #MultiscaleModelling #Mechanobiology #Microfluidics

Attending the Aarhus PtX symposium. A sneak preview of C2CAT's poster. More to come. #power2x #energystorage #cleanenergy #conference2022 #multiscalemodelling #CaaS #catalysis #computationalchemistry #synthesis #power

The final countdown! Barbara’s thesis defence is approaching! This Friday at 1015 in Pascal (DM for zoom). In the meantime, you can read her thesis and interview she gave for @Naturvetenskap ! #ageing #multiscalemodelling gu.se/en/news/mathematics-to…

J Chem. Theory Comput. Local Temporal #Acceleration Scheme to Couple Transport and #ReactionDynamics in #KineticMonteCarlo Models of #Electrochemical Systems, from Alessio Gagliardi group at Technische Universität München @JCIM_JCTC #MultiscaleModelling pubs.acs.org/doi/10.1021/acs…