You want to optimize a molecule with respect to a single or multiple properties? You can do it in just a few lines of code. Check out the implementation of our recently proposed Molecular Swarm Optimizer! github.com/jrwnter/mso

Great work by Floriane Montanari and co-workers Bayer AG. They use Multitask Graph Convolutional Networks to predict physico-chemical ADMET properties of molecules. #cheminformatics #compchem

Moritz Hoffmann figured out a way to generate valid Euclidean distance matrices without going through the 3D representation. We believe this may be a very useful tool when generating molecules or other particle data. arxiv.org/abs/1910.03131

Moritz Hoffmann published the code to generate valid Euclidean distances matrices without going through the 3D embedding. Useful for generating molecules, point clouds etc. github.com/noegroup/EDMne… Paper: arxiv.org/abs/1910.03131

"Machine learning for molecular simulation" is the #1 paper on Arxiv today in chemical physics. Congrats Frank Noe. See it at -> assert.pub/arxiv/physics/… and assert.pub/papers/1911.02…. Please retweet.

Excited to see our ChemRxiv #preprint published in Nature Communications Thanks to Benoît Baillif David Rouquié Joerg Wichard and Djork-Arné Clevert for being an amazing team at Bayer AG We combined #ArtificialIntelligence #AI with #GeneExpression for #DrugDiscovery nature.com/articles/s4146…

Stochastic Normalizing Flows: We can train normalizing flows by combining invertible networks and stochastic dynamics (MCMC, Langevin) in any order. Better expressivity and sampling than either one. With Jonas Köhler & Hao Wu #MachineLearning #DeepLearning arxiv.org/abs/2002.06707

Check out our latest work on a small web application for interactive generative molecular design, just published in Bioinformatics. Paper: doi.org/10.1093/bioinf… Source code + Docker image: github.com/jrwnter/grueni… Showcase: youtube.com/watch?v=7KAgEb…

New preprint with Tuan Le, Robin Winter and Djork-Arne Clevert: You can, with a surprising success rate, reengineer drug chemical structures from standartized fingerprints! Also: welcome Tuan to the group!! bit.ly/2YZ3adr

Happy to start my PhD at the AI4Science group from Frank Noe. Check out our recent publication on reverse-engineering molecular structures from classical fingerprints that are often used in cheminformatics. Paper: chemrxiv.org/articles/Neura…

Wohoo - PauliNet, our deep #NeuralNetwork solution of the electronic Schrödinger equation is out in Nature Chemistry. Huge respect for @jhrmnn and Zeno Schätzle for championing this. Deep Quantum Monte Carlo is the future! nature.com/articles/s4155…

1/4 Check out our recent #preprint in ChemRxiv! We applied geometric #DeepLearning to predict ligand binding conformations. Thanks to the team: Joerg Kurt Wegner Antonio del Rio Chanona and Mazen Ahmad from JNJInnovative Medicine and Imperial ChemEng Open access: chemrxiv.org/articles/prepr…

Very excited to share that our paper “Permutation-Invariant Variational Autoencoder for Graph-Level Representation Learning” has been accepted at #NeurIPS2021 ! Joint work with Frank Noe and @OkkoClev Preprint: arxiv.org/abs/2104.09856

Excited to join Microsoft Research on this journey. Big shoutout to Chris Bishop, Sebastian Nowozin, Max Welling, Rianne van den Berg, Marc Brockschmidt, Krzysztof Maziarz, Marwin Segler, Becky Frost and the whole team for being super welcoming and amazing.

My recent work on insilico protein-ligand binding prediction using unsupervised learned representations is out! Check it out here, feel free to DM me if you have questions. mdpi.com/1422-0067/22/2…

Thrilled to see our paper "A geometric #DeepLearning approach to predict binding conformations of bioactive molecules" featured on the cover of Nature Machine Intelligence Check it out: rdcu.be/cDy5f #chemtwitter #AI #MachineLearning #DrugDiscovery #CompChem #ML #Cheminformatics

We are thrilled to announce the !AI.QU. project on bridging AI & QUantum mechanics 🤖🖥️ for rapid, accurate, and scalable in silico predictions, welcoming our first partners Nong Artrith Utrecht University Faculty of Science, Utrecht University & FACCTs !AI.QU. is funded by Bayer AG's Life Science Collaboration

We propose a new+data-efficient way for #MachineLearning coarse-grained molecular force fields from simulation data (e.g. all-atom). With Jonas Köhler, Andreas Krämer, Yaoyi Chen, Cecilia Clementi. BIFOLD Microsoft Research European Research Council (ERC) Freie Universität arxiv.org/abs/2203.11167

There is now an updated version of our #NeurIPS2022 paper on arxiv. Most notably, we included new experiments were we apply our method on #molecular #conformations from QM9 (3D point cloud) and #ShapeNet (3D voxel data).