Differentiating drug response using unsupervised segmentation
Updated: Nov 17, 2018
We applied a novel segmentation algorithm to split a population of cell lines based on differences in drug response.
The effectiveness of a particular drug has predominantly been analysed in isolation and there lacks data-driven approaches to consider the full response pattern between multiple drugs to study biomarkers at the same time. To reveal subpopulations where the pharmacological response between compounds agree and diverge, we applied a novel population segmentation algorithm, POET, to compare 344 drug pairs targeting the MAPK and PI3K-AKT pathways across >800 genomically-diverse cancer cell lines. We show that POET was capable of integrating multiple measures of drug response to identify subpopulations that differentiate response to inhibitors of the same or different targets. MEK, BRAF and PI3K inhibitors with different sensitive subpopulations were shown to be effective as combined therapies, particularly when stratified for BRAF mutations. This data-driven approach paves a new way for patient stratification to identify novel cancer subpopulations, their genetic biomarkers, and effective drug combinations. Find out more here.