Visual Universitätsmedizin Mainz

Research Projects

Deciphering the epigenetic basis of resilience

This study examines the molecular mechanisms regulating neuronal function in stress susceptible and resilient mice following chronic social defeat (CSD) by comparing the transcriptomes of activated cells in selected brain regions using RNA-seq. In addition, using the assay for transposase-accessible chromatin followed by high-throughput sequencing (ATAC-seq) and bisulfite sequencing (Bis-seq), we monitor the global changes in chromatin accessibility and DNA methylation signatures that alter the functional states of neurons in response to stress.

Using Multi-Omics Integration to explore the Molecular Background of Stress Resilience

Resilience is the ability to cope with stress or to quickly recover to pre-crisis state after being exposed to extreme stress. As the human mental status is highly diverse, there are presumably many different molecular mechanisms underlying resilience. The main goal of this project is the identification of common molecular patterns between resilient individuals by analyzing and integrating various omics levels, e.g. transcriptomics, proteomics, methylomics and metagenomics. By subsequently selecting important features within the large amount of data, we aim to give a less complex view on the dynamic process of resilience. This work is funded by the Leibniz Institute for Resilience Research (LIR) ( and in collaboration with the Frauenhofer ITWM ( The majority of the underlying data sets originate from the MARP ( and LORA ( studies from the LIR.

Reliability of Next Generation Sequencing Machines and Bioinformatic Processing Pipelines

Next Generation Sequencing (NGS) is commonly used to gain insights into various questions from biology and medicine. This approach depends heavily on reliable and accurate data. In this project, we investigated the impact of different sequencing machines and their respective bioinformatic processing pipelines on the resulting variant call data.

We examined 99 individuals, which were sequenced three times with three different sequencers and were post-processed with different bioinformatic processing pipelines. The analysis showed considerable heterogeneity between the three sequencing cohorts. Even after applying non-specific filters like MAF and HWE, which are the standard for quality filtering and noise reduction in NGS cohorts, differences of several thousand variants per individual remained. This illustrates the need for continued refinement of both laboratory and computational analysis methods in order to achieve bias-free data.

BIG data integration of genetic and epigenetic variations in neurodegenerative diseases

To gain a better understanding of the global mechanisms underlying neurodegeneration we use supercomputing facilities and recently developed High-Performance Computing methods in multivariate Genome Wide Association Studies (GWAS) for the extraction of global patterns. Analysis includes genetic as well as epigenetic and transcriptional aspects, underlying neurodegenerative diseases i.e. Alzheimer’s, Parkinson’s and Huntington’s disease. Via a trans-Omics evaluation followed by in silico modeling we hope to extract core (biochemical) networks across multiple omic-layers (genome, transcriptome, methylome).

Optimization of the Calcium Imaging Analysis Pipeline

Calcium Imaging Analysis enables researchers to track the activity of hundreds and thousands of neurons within the brains of living animals. However, analyzing this huge amount of data raises some difficulties. Together with the group of Prof. Dr. Albrecht Stroh ( and in collaboration with the Fraunhofer ITWM, we are following the goal of improving and accelerating the analysis of calcium imaging data, starting with the automatic marking of neurons in the generated image files using deep learning. By optimizing and simplifying the application through user-friendly interfaces of the entire analysis pipeline, the effort and time required for Calcium Imaging Analyses can be drastically reduced and will also be less dependent on personal decisions.

Uncovering neural subpopulations and transcriptional networks underlying stress resilience

Chronic stress and traumatic events have major impact on human psychological health and the manifestation of stress-related mental disorders. The majority of individuals encountering such stress are able to surmount it and therefore are resilient, while a small portion of the population manifest mood disorders, such as anxiety, social dysfunction and depression. However, very little is known about the rewiring of gene regulatory programs underlying these processes.

The chronic social defeat (CSD) is known to evoke responses in distinct brain areas, especially hippocampus, prefrontal cortex, and amygdala. In this project, we conduct single-cell RNA-seq from the hippocampus and basolateral amygdala of socially defeated mice from resilient and non-resilient groups and use this approach not only to reveal distinct neural subpopulations in response to CSD, but also to uncover whether these subpopulations behave differently in resilient and non-resilient mice. Using extensive computational analysis, we aim at characterizing gene expression programs that operate at single cell level in response to stress and how such networks are differentially (re)programmed between resilient and non-resilient mice.

Characterization of gut microbiota composition using 16S rRNA sequencing

Sequencing of the 16S ribosomal RNA marker gene (16S rRNA) provides a cost-effective method to characterize the bacterial composition of biological or ecological samples. In experimental animal studies, this approach allows investigating the impact of factors such as pharmacological treatment or diet on commensal microbiota. Another crucial research question is whether shifts in gut bacterial composition are associated with disease phenotypes. In ongoing collaborations with experimental groups, we focus on the bioinformatical and biostatistical analysis of 16S sequencing data. This requires applying ordination techniques which can handle distance measures appropriate for ecological data such as principal coordinates analysis or correspondence analysis. Furthermore, no consensus on the optimal statistical technique for differential abundance analysis exists. Therefore, this decision needs to be based on considerations such as sample size, experimental setup and specific research question.