Institut für Medizinische Biometrie
und Statistik (IMBI)

Analysis Pipelines for High-Throughput Data

In this project, we apply and optimize comprehensive data analysis pipelines from raw data to high-level statistical analyses. We generate realistically simulated data to select the optimal analysis strategy. We also investigate the sensitivity/robustness of results to choice of algorithms and configuration parameters.

Responsible: Eva Brombacher, Eva Kohnert, Carlotta Meyring, Clemens Kreutz

Collaborators: Oliver Schilling, Fabian Cieplik

Statistical and bioinformatic analyses of proteomics and other omics data

In this project, we develop computational approaches for analysis of mass-spectrometry based proteomics data. We also benchmark approaches, in particular methods for data-independent acquisition (DIA).

Responsible: Eva Brombacher, Clemens Kreutz

Our collaborators: Oliver Schilling

Statistical and bioinformatic analyses of microbiome data

In this project, we develop and apply statistical approaches for analysis of sequencing-based microbiome data. We currently focus on statistical models that account for the compositional nature, excessive zero counts, overdispersion, and patient-specific effects.

Responsible: Eva Kohnert, Carlotta Meyring, Clemens Kreutz

Our collaborators: Ann-Kathrin Lederer, Fabian Cieplik, Ali Al Ahmad

Analysis of Single Cell Data

In this project, we compare the performance of statistical models and tests as well as machine learning methods (e.g. clustering) for processing of single cell data, e.g. from scRNA-seq experiments. We also establish pipelines which are robust against the choice of configuration parameters.

Responsible: Eva Kohnert, Carlotta Meyring, Clemens Kreutz

Our collaborators: Florens Lohrmann, Philipp Henneke

AI & Foundation Models for Single Cell Data

In this project, we compare the performance of AI-based models and tests (e.g. scGPT) for analyzing of single cell data, e.g. from scRNA-seq. We also fine-tune foundation models for downstream tasks such as cell type annotation, predicting the expression of unobserved genes or for unobserved biological conditions.

Responsible: Jonatan Menger, Eva Brombacher, Clemens Kreutz

Our collaborators: Andreas Raue, Wolfgang Schamel, Thomas Ott, Thomas Brox

Modelling Methods in Systems Biology

In this project, we develop and extend existing methodology for establishing mathematical models of biochemical processes in living cells. We focus on

• Development of reliable and robust optimization methods for parameter estimation
• Deriving low dimensional representations of the dynamics of regulation networks
• Further advancing the RTF modelling approach as an novel modelling approach
• Approximating differential equation models by the RTF for multi-scale modelling
• Evaluation of deep learning approaches in the context of ODE modelling
• Implemention of novel methods as R package or into the Data2Dynamics modelling framework

Responsible: Niklas Neubrand, Timo Rachel, Tim Litwin, Clemens Kreutz

Our collaborator: Olaf Groß, Jens Timmer, Jan Hasenauer

Benchmarking in Systems Biology

In this project, we evalute the performance of existing computational methods for mathematical modelling by

• Comparison of the performance of optimization methods which are applied for model calibration/parameter estimation
• Establishment of benchmark problems which can be used to assess modelling methods
• Development and extension of methodology for performing reliable benchmark studies
• Establishment of approaches for the realistic generation of simulation data

Link: Benchmark models on github

Responsible: Lukas Refisch, Janine Egert, Clemens Kreutz

Our collaborators: Jens Timmer, Jan Hausenauer

Modelling in Developmental Biology

In this project, develop spatio-temporal models of cellular signalling and cell type specification of Xenopus (cilia and mucociliary development).

Responsible: Tim Litwin, Clemens Kreutz

Collaborators: Peter Walentek

Data2Dynamics

In this project, we develop and extend the Data2Dynamics modelling environment which is a high-performance expert implementation for mathematical modelling.

Link: D2D repository at github

Responsible: Tim Litwin, Niklas Neubrand, Timo Rachel, Clemens Kreutz

Our collaborators: Jens Timmer, Andreas Raue