Our group's new work focusing on deep learning-based QSAR

Predicting Retrosynthetic Reaction using Self-Corrected Transformer Neural Networks by Jun Xu, Yuedong Yang & co-workers bit.ly/2XlTYvj

Deep Scaffold Hopping with Multi-modal Transformer Neural Networks by Yuedong Yang & co-workers doi.org/10.26434/chemr…

Glad to review papers in ML4H ML4Health Thanks Shameer for his patience and guidance！

Excited to share our latest dataset paper: "Reactzyme: A Benchmark for Enzyme-Reaction Prediction"! 🚀 ArXiv: arxiv.org/abs/2408.13659.

It's been a while since I last posted here—so excited to see our work gaining attention! I'll be sharing more of our latest research. DynamicBind is one of our recent projects, focused on flexible protein docking. Give it a try and have fun! github.com/luwei0917/Dyna…

Awesome!

EnzymeFlow: Reaction-conditioned Enzyme Catalytic Pocket Design 1/ 🚀 Introducing EnzymeFlow - a revolutionary generative model designed to create enzyme catalytic pockets tailored for specific reactions. This breakthrough could significantly impact biotechnology, synthetic

Retrieval Augmented Diffusion Model for Structure-informed Antibody Design and Optimization • The RADAb framework introduces a novel retrieval-augmented diffusion model to design and optimize antibodies using both structural and evolutionary information. • A key innovation

BindingGYM: A Large-Scale Mutational Dataset Toward Deciphering Protein-Protein Interactions 1. Introducing BindingGYM, the largest dataset of high-throughput mutational data for protein-protein interactions (PPIs), with over 10M raw data points distilled into 500K high-quality

EnzymeCAGE: A Geometric Foundation Model for Enzyme Retrieval with Evolutionary Insights 1. EnzymeCAGE introduces a novel framework combining enzyme structure, evolutionary insights, and chemical reactions to predict enzyme functions and retrieve enzymes for reactions. It

🧬Thrilled to share our new preprint on predicting scRNAseq after perturbations! We benchmarked 9 models (including foundation models) across 17 datasets with 24 metrics. Check out: biorxiv.org/content/10.110…. in short: Foundation model is at least better than linear baeline

GerNA-Bind: Geometric-enhanced RNA-ligand Binding Specificity Prediction with Deep Learning 1. GerNA-Bind introduces a geometric deep learning framework that models RNA-ligand interactions by incorporating multi-state RNA-ligand representations, improving binding specificity

Capturing Natural Evolution in Function-guided RNA Design via Genomic Foundation Models 1. This study introduces RILLIE, a zero-shot RNA design framework that combines large language models (LLMs) and inverse folding models (IFMs) to simulate natural evolution and optimize RNA

📢 We are excited to announce OriGene, a self-evolving virtual disease biologist poised to revolutionize therapeutic target discovery. This AI agent system autonomously discovers novel, mechanistically grounded drug targets at scale, addressing the critical bottleneck in drug

Accurate Protein-Protein Interactions Modeling through Physics-informed Geometric Invariant Learning １．ProTact is a new SE(3)-invariant geometric graph neural network designed to predict inter-protein contacts with high accuracy, especially in cases where evolutionary data is

CAMEL-AI now integrates Origene Toolkit! 🔬 Key features: ✔️ 640+ integrated tools and databases (ChEMBL, PubChem, UniProt) ✔️ AI-powered multi-agent system for automated therapeutic discovery ✔️ Proven workflows for molecular, chemical, and disease research With Origene,

Fitness Aligned Structural Modeling Enables Scalable Virtual Screening with AuroBind 1. AuroBind is a novel virtual screening framework that integrates atomic-level structural modeling with binding fitness prediction, achieving state-of-the-art performance in identifying

📢 New preprint: AuroBind, a structure foundation model fine-tuned for high-throughput virtual screening, with experimental validation across diverse targets. It provides a scalable solution for one-shot small molecule hit identification, enabling ultra-high-throughput screening

🔥We’re excited to announce ODesign The first general-purpose molecular design world model, built by Lingang Lab, Shanghai AI Lab, CUHK, SJTU, ZJU, UW, and Harvard. 🧬 ODesign designs everything: proteins, peptides, RNAs, DNAs, small molecules, and even metal ions — for any