Congratulations to Xinpo Lou on the publication of his two recent works in Journal of Chemical Information and Modeling and Journal of Cheminformatics!

The proposed MambaTransDTA model introduces a novel hybrid Mamba–Transformer architecture for drug–target binding affinity prediction. By effectively combining long-range dependency modeling with local interaction learning, the model achieves strong and consistent performance improvements across multiple benchmark datasets, providing a valuable contribution to AI-driven drug discovery. For more details, please read the full article at https://doi.org/10.1021/acs.jcim.5c02361.
#Layers #Molecular Modeling #Molecules #Peptides And Proteins #
Screening Assays

In parallel, the proposed Multi-MoleScale framework introduces an innovative multi-scale learning strategy that integrates graph contrastive learning with sequence based representation learning, providing a powerful advance for molecular property prediction. By jointly modeling molecular structural and contextual information, the framework achieves strong performance across multiple benchmark datasets. The study showcases the strong potential of multi-scale, self-supervised learning techniques for advancing molecular property prediction in AI-driven drug discovery. The full article is available at https://doi.org/10.1186/s13321-025-01126-w.
#GCL #Molecular Property Prediction #Multi-MoleScale #Deep learning #Self-supervised learning