Distinguished Speaker Series: A Tale of two tales: Studying (a) the CRISPR risk of mutation selection and (b) extracting cell-type information from bulk expression
Prof. Eytan Ruppin, Chief, Cancer Data Science Lab, NCI, NIH
Abstract:
(a) Studying CRISPR risk of mutation selection: Analyzing genome-wide CRISPR and RNAi screens we find that CRISPR gene editing can select for KRAS and VHL mutants, at a level reported experimentally previously for p53. These predictions are further by analyzing independent CRISPR screens and patients’ tumor data and via a new set of CRISPR screens. These results point to the potential selection of specific cancer driver mutations during CRISPR-Cas9 gene editing (see https://www.biorxiv.org/content/10.1101/407767v2).
(b) Extracting cell-type information from bulk expression: We developed a novel transcriptomics deconvolution computational tool termed CompuSort. Given patients’ tumor bulk expression data, CompuSort estimates both cell type abundance and mean cell type gene expression levels for each patient individually. Analyzing 10,000 TCGA tumor samples we identify ligand-receptor interactions between tumor and immune cells that are associated with patients’ survival. Deconvolving expression data of immune checkpoint blockade treatment in melanoma patients, we build a robust predictor of response to therapy that outperforms existing ones. The future application of CompuSort may facilitate the study of cancer and other complex disorders.
Host: Prof. Ron Shamir, School of Computer Science
