Network-based multi-drug response prediction using multi-omics data

Pauline Hiort

Data Analytics and Computational Statistics
Hasso Plattner Institute

Office: HPI Campus I, K-E.17
Tel.: +49 (0)331 5509 - 4984
Email: pauline.hiort(at)hpi.de

Supervisor: Prof. Dr. Bernhard Renard

With our research we focus on multi-drug response prediction using multi-omics networks. Our goal is to develop a pipeline to predict the response to combinations of drugs in diseases such as cancer.

Research Topic

Biological systems are very complex, with thousands of genes, proteins, and other molecules interacting in a directed or undirected manner. These systems can be modeled with networks [1]. Networks allow for visualizing and analyzing interactions between entities in complex systems, e.g., people in social networks or genes in biological networks. Networks are modeled as graphs with a given set of nodes and edges. The nodes represent the investigated entities, while the edges represent the relationships between these entities.

Using networks, different types of analyses of biological systems can be done, such as predicting the system’s response to a drug [2, 3, 4]. Testing drugs and their combination with patients or in a lab is costly and time-consuming. Therefore, computational testing of these drugs and consequently reducing the number of potential drugs for a specific disease is beneficial for medical research. The administration of drugs causes perturbations of the underlying system that can be analyzed using a representative biological network. For these analyses, several different information layers, i.e., omics (genomics, proteomics, and metabolomics) levels, can be utilized. Thus, multi-omics networks combine multiple layers of biological information like gene expression levels, protein quantities, and other available biological information. We are developing pipelines to predict the responses to (multiple) drugs using networks from multi-omics data.

Research Projects

DrDimont – Drug response predictions from differential analysis of multi-omics networks – workflow. Figure adapted from Hiort et al. [5].

DrDimont: Explainable drug response prediction from differential analysis of multi-omics networks [5]:

We developed a novel pipeline, DrDimont, that predicts the differential drug response of two different phenotypes, e.g., cancer subtypes, employing multi-omics networks [5]. Our biological networks include data from gene expression analyses, protein quantifications, and metabolite quantifications. With DrDimont, two multi-layer networks are constructed for the two different conditions based on the correlation of the respective biological omics level. After combining the two networks to a differential network, the drug response is predicted. The differential drug response indicates whether a drug might have a different effect in the two groups. As a case study, we analyze the differential drug response in breast cancer data of ER+ (Estrogen Receptor positive) and ER- (Estrogen Receptor negative) patients. The tool DrDimont is available on CRAN and results have been published here.

Exemplary drug combinations in a protein-protein interaction network

Network-based drug-drug combination response prediction using multi-omics data:

We are developing a pipeline for drug-drug combination classification employing multi-omics networks. Cheng and Kovacs et al. [4] developed a workflow for distinguishing beneficia vs. adverse drug-drug combinations using unweighted human protein-protein interaction networks. These networks generally contain interactions measured in the healthy system. We are working on extending the approach of Cheng and Kovacs et al. for application on weighted and condition-specific, data-derived networks, i.e., networks built from omics measurements of a condition, such as cancers.

Publications and Presentations

Hiort, P. et al. (2022). DrDimont: Explainable drug response prediction from differential analysis of multi-omics networks. Bioinformatics, 38, ii113–ii119. https://doi.org/10.1093/bioinformatics/btac477
Poster at European Conference on Computational Biology 2022: Hiort, P., Hugo, J., Zeinert, J., Müller, N., Kashyap, S., Rajapakse, J.C., Azuaje, F., Renard, B.Y., Baum, K. DrDimont: Explainable drug response prediction from differential analysis of multi-omics networks
Poster at German Conference on Bioinformatics 2022: Hiort, P., Hugo, J., Zeinert, J., Müller, N., Kashyap, S., Rajapakse, J.C., Azuaje, F., Renard, B.Y., Baum, K. Finding relevant data layers in multi-omics network-based drug response predictions

Teaching

Winter Semester 2021/2022

Implementing information flow in complex networks (project seminar)
Introduction to programming (Schülerkolleg 7^th-8^th grade)

References

Ramos, P., Arge, L., Lima, N., Fukutani, K. F., & de Queiroz, A. (2019). Leveraging User-Friendly Network Approaches to Extract Knowledge From High-Throughput Omics Datasets. Frontiers in genetics, 10, 1120. https://doi.org/10.3389/fgene.2019.01120
Iorio, F., Knijnenburg, T. A., Vis, D. J., Bignell, G. R., Menden, M. P., et al. (2016). A Landscape of Pharmacogenomic Interactions in Cancer. Cell, 166(3), 740–754. https://doi.org/10.1016/j.cell.2016.06.017
Menden, M. P., Wang, D., Mason, M. J., Szalai, B., Bulusu, K. C., Guan, Y., et al. (2019). Community assessment to advance computational prediction of cancer drug combinations in a pharmacogenomic screen. Nature communications, 10(1), 2674. https://doi.org/10.1038/s41467-019-09799-2
Cheng, F., Kovács, I. A., & Barabási, A. L. (2019). Network-based prediction of drug combinations. Nature communications, 10(1), 1197. https://doi.org/10.1038/s41467-019-09186-x
Hiort, P. et al. (2022). DrDimont: Explainable drug response prediction from differential analysis of multi-omics networks. Bioinformatics, 38, ii113–ii119. https://doi.org/10.1093/bioinformatics/btac477