Back

Single-Cell Pharmacogenomic Landscapes of Epigenetic Therapy in Gastric Cancer

Xu, C.; Ma, H.; Sheng, T.; Lee, M. H.; Abdul Ghani, S. A. B.; Perlaza-Jimenez, L.; Huang, K. K.; Lim, S. K.; Srivastava, S.; Ong, X.; Tay, S. T.; Ho, S. W. T.; Tan, A. L. K.; Zhu, F.; Nasca, V.; Randon, G.; Ashktorab, H.; Cheng, A. S.-L.; Jeyasekharan, A. D.; Li, S.; Teh, M.; Sundar, R.; Powell, D. R.; Rosenbluh, J.; Pietrantonio, F.; Yong, W. P.; So, J. B.-Y.; Tan, P.

2025-04-21 cancer biology
10.1101/2025.04.16.649236 bioRxiv
Show abstract

Gene expression signatures ("molecular phenotypes") are extensively utilized in cancer research. To study how gastric cancer (GC) molecular phenotypes are shaped by cell-intrinsic genetic alterations interacting with cell-extrinsic therapeutic pressures, we performed direct capture Perturb-seq (dcPerturb-seq) to interrogate >200 GC-related genes across 4 distinct epigenetic drug classes in multiple gastric lines. We captured 17.7 million pharmocogenomic expression interactions in 625,866 cells representing baseline and post-therapeutic molecular phenotypes. This single-cell pharmacogenomic compendium confirmed previously known gene-driven molecular phenotypes, elucidated poorly characterized genes, and uncovered novel gene dosage-molecular phenotype relationships. Molecular phenotypes in post-therapeutic surviving cells revealed diverse gene perturbation-associated pathways causing convergent drug resistance (EMT plasticity, cell cycle alterations, metabolic reprogramming), highlighting combinatorial strategies for restoring sensitivity. Mapping of in vitro molecular phenotypes to primary human GCs imparted prognostic information and insights into spatial heterogeneity. Comparative analysis of gene perturbations across therapies and lines revealed both conserved and context-specific molecular alterations. Our results illustrate how Perturb-seq approaches can systematically map diverse cancer-associated molecular phenotypes across multiple gene/drug/cell line interactions, yielding translational insights.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.