gms | German Medical Science

Introduction: While it has been well established that drugs affect and help patients differently, personalized drug response predictions remain challenging. Solutions based on single omics measurements have been proposed, and networks provide means to incorporate molecular interactions into reasoning. However, how to integrate the wealth of information contained in multiple omics layers still poses a complex problem.

Methods: We present DrDimont, Drug response prediction from Differential analysis of multi-omics networks. It allows for comparative conclusions between two conditions and translates them into differential drug response predictions. DrDimont focuses on molecular interactions. It establishes condition-specific networks from correlation within an omics layer that are then reduced and combined into heterogeneous, multi-omics molecular networks. A novel semi-local, path-based integration step ensures integrative conclusions. Differential predictions are derived from comparing the condition-specific integrated networks. DrDimont’s predictions are explainable, i.e., molecular differences that are the source of high differential drug scores can be retrieved.

Results: Using DrDimont, we predict differential drug response in breast cancer using transcriptomics, proteomics, phosphosite, and metabolomics measurements and contrast estrogen receptor positive and receptor negative patients. DrDimont performs better than drug prediction based on differential protein expression or PageRank when evaluating it on ground truth data from cancer cell lines. We find proteomic and phosphosite layers to carry most information for distinguishing drug response, while metabolomics data seems to confound results. We show exemplarily for the drug dinaciclib which of its drug targets are responsible for differential drug response, and which types of interactions are particularly affected.

Discussion: Our findings on the relevance of proteins and phosphosites for drug response predictions seem reasonable because drugs mainly act on proteins where they interfere with their cellular functions that are frequently modulated by post-translational modifications such as phosphorylations. Differential drug response prediction is a difficult problem, and we have to rely on ground truth from cell line measurements as surrogate here. However, with its differential approach, DrDimont enables predictions also in settings where the response to a drug is less well characterized. The established combined molecular networks can be retrieved and employed for a user’s own analysis approaches.

Conclusion: DrDimont is a flexible tool for subgroup-specific and comparative predictions with an explainable framework, and we envision that it contributes with its proof-of-principle to improving the clinical decision process in the future. It is available as an R package: https://gitlab.com/PHiort/DrDimont.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.

Further presentation/publication of this contribution: Hiort P, Hugo J, Zeinert J, Müller N, Kashyap S, Rajapakse JC, et al. DrDimont: Explainable drug response prediction from differential analysis of multi-omics networks. Submitted to the Proceedings of ECCB 2022, publication (if accepted) at Bioinformatics.