Computational Discovery Research

Aim: systems biology of metabolic disorders

  • CoNI - Correlation based Network Integration

We developed CoNI, as an unsupervised integration method for numerical omics datasets. CoNI based on partial correlations to identify putative confounding variables for a set of paired dependent variables. It returns an integrated hypergraph with the nodes as the dependent variables formed by a vertex dataset and the weighted edges formed by the confounders of a linker dataset. This graph can be further analyzed and visualized for the identification of priority candidates of biological importance.

Klaus, VS., Schriever SC., Monroy-Kuhn JM. et al. (2021): Correlation guided Network Integration (CoNI) reveals novel genes affecting hepatic metabolism. MolMet

  • An Atlas of Circadian Metabolism

As part of a large-scale study, we constructed 24-hour metabolic profiles of mouse tissues and organs under conditions of energy balance and high-fat diet. Our findings provide an overview of how the various metabolic pathways in the body are interconnected and also reveal suitable time frames for anti-obesity therapies.

 

Dyar, KA. & Lutter, D. et al. (2018): Atlas of Circadian Metabolism Reveals System-wide Coordination and Communication between Clocks. Cell;

  • Calimera - CALorimetry Improved MEtabolic & Respiratory data Analysis

We develop a framework to automatically process and analyze multiple features derived from indirect calorimetry data including additional parameters such as physical activity or body weight. We basically use splines to estimate underlying metabolic functions for individual parameters. This does not only allow us to apply novel statistical tests to identify differences in circadian oscillations, but also allows for extraction of different features from the data that we can directly compare between individuals, dietary conditions or different mouse strains.