Galadriel Brière

This study evaluates how data leakage affects KGE models' performance in biomedical link prediction and finds that current benchmarks may overestimate generalization.

PhenoXtract extracts phenotypes from clinical descriptions using a hybrid approach of Large Language Models and Knowledge Graph embeddings, achieving high recall and precision.

The paper develops a network-based method to identify novel phenotypes for rare diseases by integrating clinical data with prior knowledge from Orphanet and HPO.

The paper demonstrates the versatility of MultiXrank, a Random Walk with Restart algorithm on multilayer networks, for gene prioritization, gene-disease association prediction, and disease signature computation.

The paper introduces ClustOmics, a consensus clustering tool for multi-omic data integration, showing it produces robust consensus clusters across ten cancer types.