We develop a wide variety of statistical and data science methods and workflows for biological (microbial) data across different biological time
and spatial scales.
Compositional changes of cell types are main drivers of biological processes.
However, their detection through single-cell experiments is difficult due to the compositionality of the data and
low sample sizes.
scCODA, a Bayesian model addressing these issues, enables the study of complex cell type effects in disease,
and other stimuli. scCODA demonstrates excellent detection performance and could identify experimentally verified
cell type changes that were missed in original analyses.
The method was developed by Johannes in collaboration with the groups of
Benjamin Schubert and
You can read more about scCODA in Nature Communications and readthedocs.
In this project, we aim to identify stressors and bacterial regulatory pathways that control host adaptation and antibiotic sensitivity. By using a high-throughput chemical genomics approach, we will explore chemical stimuli from the host environment that induce stress response pathways in Salmonella and Campylobacter, and elucidate the underlying molecular crosstalk between sensory pathways of these microbes.
This project is part of the bavarian Bayresq research consortium, and is done in close colaboration with members of
Dr. Cynthia Sharma's and Dr. Ana Rita Brochado's labs. For more info on the role of our group in this project you can contact Roberto
In this project, we combine chemical biology, biochemical, and proteomic approaches with cell biology and computational analyses to unravel how epigenetic effector molecules can read DNA and histone modification patterns and how they recognise different chromatin modification states. Our goal is to decipher the “epigenetic code” by identifying epigenetic reader molecules that can integrate information from multiple chromatin modifications and to understand how these factors operate at the molecular level both in healthy and pathological conditions.
This project is done in close collaboration with Dr. Till Bartke from the Institute of Functional Epigenetics at Helmholtz Zentrum München. In our group Mara is responsible for this project.
NetCoMi is an R package that provides an extensive toolbox for constructing, analyzing, and comparing networks for microbial compositional data.
It also offers the ability to quantify network differences and thus enables insights into whether single taxa, groups of taxa, or the overall network structure change between groups.
The package was developed by Stefanie as part of her Ph.D. thesis, in collaboration with her supervisors and the group of
Prof.Dr. Erika von Mutius at Helmholtz Zentrum München.
The manuscript describing the workflow and theory behind the package is published
in Briefings in Bioinformatics.
Microbial or cell type compositions from high-throughput sequencing experiments are often ordered by a tree structure, such as taxonomic or cell lineage trees.
The tascCODA method, which was developed by Johannes,
utilizes this structure to dynamically infer compositional changes on subgroups of the composition.
Compared with other methods, tascCODA has very good detection accuracy and gives a more parsimonious picture of the compositional shifts in the data.
The method was published in Frontiers in Genetics, and the Python implementation can be found on our github page.
In this project, we study food allergy and non-IgE microbial
associations in a human gut. The investigation relies on
computational techniques based on high-dimensional statistics,
including sparse regression, graphical model techniques, and deep
learning approaches. Also, we simulate a synthetic cohort capturing
the underlying correlation structure of KORA - a large-scale private
population-based cohort with thousands of participants. We contribute
with a public microbiome cohort at an unprecedented scale together
with causal inference methods which discover taxon-disease
associations and are reproducible in open-source software