Jérôme Tubiana

The paper develops cell-permeable C16orf74 peptides that disrupt Calcineurin-NFAT interaction, inhibit T-cell activation, and show minimal cytotoxicity compared to Cyclosporine A.

LocAlign uses geometric deep learning to align protein structures and identify functional motifs, achieving high accuracy for both similar and dissimilar folds.

The paper uses a generative model to uncover neural assemblies in zebrafish brains, showing it can replicate activity statistics and reveal functional connectivity.

ScanNet predicts protein binding sites using an interpretable geometric deep learning model, achieving accuracy on unseen protein folds and identifying epitopes in SARS-CoV-2 spike protein.

This paper introduces a flexible machine learning method for predicting antigen presentation by HLA-I alleles and shows its improved accuracy for rare alleles.

This paper introduces a blind sparse deconvolution algorithm for inferring spike trains from fluorescence recordings, showing improved robustness, speed, and accuracy over existing methods.