Preview

PDF, English - main document Download (7MB) | Lizenz: From Activity Inference to Multi-Omics Network Contextualization: Deciphering Cellular Signaling and Disease Mechanisms by Müller-Dott, Sophia underlies the terms of Creative Commons Attribution 4.0

Abstract

The integration of diverse omics layers with advanced computational methods can help to decipher cellular signaling and disease mechanisms. Thereby it is crucial to ensure that computational predictions truly reflect biological mechanisms and that different omics layers are cohesively integrated. This thesis focuses on evaluating approaches to infer the activity of transcription factors and kinases as well as advancing methods to uncover context-dependent signaling networks. First, to identify the most reliable strategies for activity inference, benchmarking frameworks were established to assess various inference methods. This revealed that a novel collection of signed transcription factor-gene interactions outperforms existing resources in predicting transcription factor activities. Similarly, manually curated kinase-substrate libraries combined with less complex computational models were shown to provide higher accuracy for kinase activity inference. Next, to reveal the role of these regulators in signaling pathways across diverse biological contexts, methods for network contextualization were developed, incorporating phosphoproteomics data alone and in combination with transcriptomics data. For phosphoproteomics-based network modeling, signed protein-protein interactions were incorporated to account for regulatory directionality, improving the representation of biological networks. Additionally, a multi-omics network contextualization approach was established which is able to link upstream stimuli to kinase and transcription factor activities in a cohesive manner, bridging phosphoproteomics and transcriptomics data. The network models were then applied to study the effects of metformin on colorectal cancer and the mechanisms driving hepatic stellate cell activation, uncovering condition-specific regulatory mechanisms and potential interactions between key signaling pathways. This highlights that integrating experimental data with reliable prior knowledge and advanced computational approaches can aid in understanding context-dependent signaling processes in complex biological systems.