Georgios Karakatsoulis

The paper develops synth4bench, a synthetic data pipeline to benchmark tumor-only somatic variant callers, revealing inconsistencies in tool performance and highlighting the need for sequencing optimization and caller selection.

This study identifies lineage-associated tumor cell states in pediatric osteosarcoma that correlate with clinical outcomes and guide immunotherapeutic target selection.

APNet identifies differentially active drivers of severe COVID-19 using a sparse deep learning model combined with co-expression networks, outperforming existing methods in patient classification.