Hebrew University of Jerusalem

Protein sequence embedding model, contrastively fine-tuned from ESM-2, that places functionally and structurally related proteins close together.

Contrastive dual-encoder model embedding protein domains and peptides in one space to predict domain-peptide binding specificity at proteome scale.

Nucleotide language model for short metagenomic reads, assigning taxonomic domain, coding potential, and reading frame from reads down to 100 bp.

Flexible protein-ligand docking and binding affinity prediction from an apo receptor structure and ligand SMILES, using an 8-layer pair transformer.

Graph neural network that predicts magnesium ion binding sites on RNA structures, powering SAXS-based validation of RNA solution conformations.

Spatial transcriptomics prediction from H&E slides, inferring spot-level expression and tumor microenvironment composition in breast cancer.

Protein language model pretrained on UniRef90 with masked language modeling and Gene Ontology annotation prediction, at 16 million parameters.