The 6th Imric Retreat

Connecting tumor genomics and transcriptional states across from single cell measurement

Matan Hofree
The School of Computer Science and Engineering, The Lautenberg Center for Immunology and Cancer Research, The Hebrew University of Jerusalem, Israel

A central goal in cancer biology is to connect an individual tumor`s somatic genome to the transcriptional states of tumor and host cells. However, at single-cell resolution both are obscured by limitations in measurement technique and experimental confounders. Somatic mutations often appear functionally neutral, and single-cell RNA sequencing (scRNA-seq) captures only a sparse, expression-biased view of each cell`s genome. At the same time, the rare, clinically relevant cell states of greatest interest are buried beneath dominant cell-type and inter-patient variation. Our lab develops computational frameworks to uncover biologically meaningful signal on both axes from the same tumor samples. On the genomic axis, we integrate patient-matched whole-exome sequencing with scRNA-seq to prioritize transcript-supported mutations, identify recurrent hotspots and track inherited variant sites to map allelic loss – separating large chromosomal alterations from focal, copy-number-neutral events. On the transcriptional axis, we are developing NetCellSig, a novel framework that uses network propagation to amplify weak expression programs in a targeted manner, allowing us to resolve rare subpopulations of interest. Using this approach we recover an epithelial–mesenchymal transition signature with markedly improved sensitivity over standard references and nominate previously unrecognized regulators of tumor-cell plasticity. Together, these approaches outline a strategy for characterizing principles of tumor evolution – from rare genomic drivers to transcriptional plasticity and mesenchymal transition – directly from single-cell data. These methods and findings propose new ways to interpret progression and predict treatment response across cancers.