Eyal Akiva

The paper introduces PiNUI, a PPI dataset designed for ML with improved representativeness, and shows it outperforms existing datasets in PPI tasks.

The paper improves protein subcellular localization prediction by combining sequence-based language models and structure-based graph neural networks, achieving better performance with lower computational cost.

The paper introduces an unsupervised method using Transformer models, UMAP, and spectral clustering to organize protein sequences into biologically meaningful clusters.