Network- and enrichment-based inference of phenotypes and targets from large-scale disease maps

Abstract

Complex diseases are inherently multifaceted, and the associated data are often heterogeneous, making linking interactions across genes, metabolites, RNA, proteins, cellular functions, and clinically relevant phenotypes a high-priority challenge. Disease maps have emerged as knowledge bases that capture molecular interactions, disease-related processes, and disease phenotypes with standardized representations in large-scale molecular interaction maps. Various tools are available for disease map analysis, but an intuitive solution to perform in silico experiments on the maps in a wide range of contexts and analyze high-dimensional data is currently missing. To this end, we introduce a two-dimensional enrichment analysis (2DEA) approach to infer downstream and upstream elements through the statistical association of network topology parameters and fold changes from molecular perturbations. We implemented our approach in a plugin suite for the MINERVA platform, providing an environment where experimental data can be mapped onto a disease map and predict potential regulatory interactions through an intuitive graphical user interface. We show several workflows using this approach and analyze two RNA-seq datasets in the Atlas of Inflammation Resolution (AIR) to identify enriched downstream processes and upstream transcription factors. Our work improves the usability of disease maps and increases their functionality by facilitating multi-omics data integration and exploration.

Network analyses reveal new insights into the effect of multicomponent Tr14 compared to single-component diclofenac in an acute inflammation model

“…Using a systems biology approach for the AIR, we developed two plugins (Omics and Xplore) comprising various tools to integrate and analyze multi-omics data and explore the role of feedback mechanisms in molecular interaction networks [ 71 ]. The tools enable in silico perturbations and network-based enrichment experiments to identify regulated immunological phenotypes and processes.…”

“…Further, we ranked the regulators of each phenotype by their expression value and influence score. The methodology underlying the plugin is described in detail in our recent work [ 71 ]. The analysis was complemented by a literature search for additional evidence on the highest-ranked regulators in the context of acute inflammation and resolution.…”

A proteomic profile of the healthy human placenta

“…The NaviCenta consists of human and machine-readable representations of processes involved in healthy and dysfunctional placentae in SBGN process description and activity flow ( https://www.sbi.uni-rostock.de/minerva/index.xhtml?id=NaviCenta ). Phenotype predictions based on user defined data uploaded onto the NaviCenta were performed with plugins [ 23 ]. The uploaded information consisted of HGNC identifiers for the genes of all identified proteins, their difference score as calculated in Perseus, and adjusted p values.…”

Cell-Type-Specific Gene Regulatory Networks of Pro-Inflammatory and Pro-Resolving Lipid Mediator Biosynthesis in the Immune System

“…For each LM class p, we calculated a weighting factor for all elements in the submaps representing their topological inclusion in the paths connected to p. We recently described this weighting approach [25]. In summary, the weighting of an element e is calculated based on the percentage of elements and paths connected to p. N paths is the number of all paths to p and N paths e ⊂ N paths are paths that go through e. N nodes is the number of elements connected to p and N nodes e ⊂ N nodes the number of elements on the path from e to p: w e,p = r SP e,p We generated a regulatory score s for each gene in G c , representing its association to LM synthesis.…”

“…We have identified key processes at each stage of inflammation and developed a standardized representation of the associated molecular interactions in so-called standardized molecular interaction maps (MIMs). The manually curated causal interactions enable the use of systems biology approaches to infer regulatory circuits, predict signal transduction pathways, or perform perturbation experiments [25]. Among others, the AIR provides a detailed description of the biosynthetic pathways of PIMs and SPMs from their precursors AA, DHA, and EPA.…”