Cell-specific prediction and application of drug-induced gene expression profiles

Abstract

Gene expression profiling of in vitro drug perturbations is useful for many biomedical discovery applications including drug repurposing and elucidation of drug mechanisms. However, limited data availability across cell types has hindered our capacity to leverage or explore the cellspecificity of these perturbations. While recent efforts have generated a large number of drug perturbation profiles across a variety of human cell types, many gaps remain in this combinatorial drug-cell space. Hence, we asked whether it is possible to fill these gaps by predicting cell-specific drug perturbation profiles using available expression data from related conditions--i.e. from other drugs and cell types. We developed a computational framework that first arranges existing profiles into a three-dimensional array (or tensor) indexed by drugs, genes, and cell types, and then uses either local (nearest-neighbors) or global (tensor completion) information to predict unmeasured profiles. We evaluate prediction accuracy using a variety of metrics, and find that the two methods have complementary performance, each superior in different regions in the drug-cell space. Predictions achieve correlations of 0.68 with true values, and maintain accurate differentially expressed genes (AUC 0.81). Finally, we demonstrate that the predicted profiles add value for making downstream associations with drug targets and therapeutic classes.

Signatures of cell death and proliferation in perturbation transcriptomics data—from confounding factor to effective prediction

“…The application can be explored with an example dataset coming from (53). Furthermore, recent methods have been developed to predict unmeasured perturbation gene expression (54) for which our cell viability prediction methods could also potentially be utilized. Also, our results suggest that removing cell death correlating genes from the gene expression signature can help to better interpret the similarities between signatures and identify mechanism of action.…”

DREIMT: a drug repositioning database and prioritization tool for immunomodulation

“…DREIMT also includes a Drug Specificity Score (DSS) that summarises the cell-specificity of a given drug across multiple cancer cell lines (Hodos et al, 2018;Suppl. Materials S4-6).…”

Applications of Genetics, Genomics and Bioinformatics in Drug Discovery

“…Although a large number of drugs have been profiled in many cancer cell lines of various tissue origins, there are still substantial missing drug-cell line combinations in these data sources. Hodos et al 12 attempted to fill the gaps by predicting cell specific drug perturbation expression profiles. The authors developed a computational framework to first arrange existing gene expression profiles into a three-dimensional array (or tensor) indexed by drugs, genes, and cell types, and then use either local (nearest-neighbors) or global (tensor completion) information to predict unmeasured profiles.…”