1/8 Just launched: DeepMolecules – a webserver for predicting protein-small molecule interactions using AI, now out in Nucleic Acids Research!.@HHU_de

RT @ItaiYanai: 🔥The "Postdoc Night Science Germany" club will have its first meeting this Friday, May 16 at HHU Düsseldorf. Join us postdoc….

RT @SGRodriques: Today, we are launching the first publicly available AI Scientist, via the FutureHouse Platform. Our AI Scientist agents….

8/8 Whether you're engineering a metabolic pathway or screening drug candidates, DeepMolecules helps you move faster – with predictions you can trust. Try it now:.Published in the NAR Web Server Issue:.

7/8 Plus: you get access to curated experimental data – so you can explore known interactions, compare with your predictions, and generate new hypotheses.

6/8 Inputs:. • Protein sequence (FASTA or plain text). • Small molecule (SMILES, InChI, KEGG ID).Outputs:. • Predicted substrates. • Kinetic parameters (kcat, KM). • Transporter specificity

5/8 No structure? No problem. DeepMolecules works purely from sequence + molecule input. Paste amino acid sequences + SMILEs – and go. You can also upload Excel files for batch predictions.

4/8 What’s under the hood?.• Transformer-based embeddings for proteins.• Numerical encodings for small molecules (SMILES, InChI, KEGG).• Gradient-boosted decision trees for prediction.• High accuracy across diverse datasets

3/8 That’s where DeepMolecules comes in. It brings together 4 AI models trained on large-scale data with deep protein & molecule representations, predicting:.• Enzyme–substrate pairs (ESP).• kcat (TurNuP).• Michaelis constants (KM).• Transporter–substrate pairs (SPOT)

2/8 Why does it matter?.Enzyme-substrate interactions, transport specificity, and kinetic parameters like kcat and KM are central to drug discovery, metabolic engineering, and synthetic biology. But they’re hard (and slow) to determine experimentally.

RT @ItaiYanai: @nightsciencepod @Yale Apple: Spotify:

RT @nightsciencepod: New Night Science Podcast episode! Martin Schwartz from @Yale talks with us about the importance of stupidity in scien….

RT @ItaiYanai: 🔥New Night Science paper!!.Discovery happens when your initial plans fall apart but it requires you to have a particular min….

RT @nightsciencepod: New Night Science Podcast episode! We talk with Ethan Mollick – bestselling author of “Co-Intelligence” – about the ne….

RT @ItaiYanai: How can we exploit AI's hallucinations to make us more creative by generating new connections and ideas? .We talk about it o….

RT @ItaiYanai: Rather than keeping our techniques private, how can the scientific community promote more transparency and open sharing for….

RT @alfonsosoler79: We're about to upload to @biorxivpreprint a preprint on the experimental relocation of RNA Polymerase genes in wich we….

RT @alfonsosoler79: I was waited two years to see this out!! The genomic location of bacterial genes is not random. Congratulation on the….

RT @ItaiYanai: Is the genome just a bag of genes? A new paper in @ScienceMagazine now reports that for two thirds of an organisms' genes th….

This was Xiao-Pan Hu’s idea, building on his earlier work on RNA gene positions:.🔗 He tested his new hypothesis brilliantly, with additional evolutionary modeling from @KJFBartoszek and @bayubeta, our wonderful collaborators at Linköping University.