Computational modeling integrating transcriptomic and vulnerability responses can predict suppressors of cell death as candidate targets for cancer therapy

Identification of novel target genes for cancer therapy is a significant challenge of biomedical research. Here, we describe a computational pipeline, which integrates transcriptomic and vulnerability responses to cell-death inducing drugs, to predict repressors of cell-death as candidate targets for cancer therapy. The candidate target genes were predicted based on two modules: the transcriptomic similarity and the correlation modules. The transcriptomic similarity module identified genes whose targeting results in similar transcriptomic responses of the death-inducing drugs, while the correlation module identified candidate genes whose expression was correlated to the vulnerability to the death-inducing drugs. The combined predictors generated by these two modules were integrated into a single ranked metric. As a proof-of-concept, we selected ferroptosis inducers as death-inducing drugs, and triple negative breast cancer as a cancer model. The pipeline could predict candidate genes as ferroptosis repressors, as demonstrated by computational and experimental validation, including experimental data of 9 representative genes, thus, highlighting the robustness and power of this pipeline. The described pipeline can be used to identify repressors of different cell-death pathways as potential therapeutic targets for various cancer types. Overall design: In this study, 3 triple negative breast cancer lines (HS578, BT549, SUM159) were treated with the ferroptosis inducers erastin and RSL3, with DMSO serving as vehicle control. Experiment was done in duplicates. RNA was extracted, and RNAseq was performed in order to study the transcriptomal response to the ferroptosis inducers.