Bioinformatic and mathematical modelling

The detection of certain mutations in cancer-associated genes is a prerequisite for the selection of many targeted therapies. The FDA-approved »MSK-IMPACT« assay for tissue samples detects more than 400 cancer-associated genes and thus offers a valuable tool for therapy decisions.

Disseminated and circulating tumor cells are the cells of origin of metastases and often differ significantly from the cells of the primary tumor. It is therefore important to isolate individual cells and test them for mutations to diagnose and monitor the progression of the disease. As this is not possible with the original IMPACT assay, we at Fraunhofer ITEM in Regensburg, together with the University of Regensburg, have adapted the wet lab and bioinformatics methods for single cells and expanded it to include additional target genes and pharmacogenetic markers.

Gene expression profiles can be used to describe the type and condition of a cell. Knowledge of the relative activity of signaling pathways derived from these profiles can also provide insights into the molecular mechanisms underlying the individual course of the disease or the effect of drugs. Accordingly, the analysis of gene expression is an important method in the development of personalized tumor therapies. To achieve a meaningful evaluation of RNA sequencing data, we have a broad spectrum of specialized bioinformatic methods at our disposal. In particular, the experimental conditions of single-cell analysis involve special requirements for data analysis that are often not covered by standard procedures and must therefore be considered individually.

For medical research, the evaluation of microscopy images is essential to characterize the phenotype, proliferation or death rate, as well as the migration behavior of cells. We analyze brightfield and fluorescence imaging data of cell cultures in 2D and 3D, such as those collected by our High Throughput / High Content Screening unit. This also includes cell painting. This method provides a comprehensive description of the cellular phenotype and allows us to detect even minute changes caused by low doses of toxic substances or promising drug candidates. Another focus is the detailed tracing of cells in video sequences (cell tracking) for the analysis of cell behavior that helps to unveil cancer characteristics important for therapy decisions.

Adapted to individual project requirements, we implement image processing pipelines that use both classic and modern methods such as neural networks, automatically assess the quality of results and interactively evaluate the various readouts extracted from the images.

During tumor growth, the unstable genome of the tumor cells is constantly altered. Detected DNA copy number variations (CNV) are therefore regarded as a molecular indicator of the malignancy of the cells and in many cases already enable an assessment of the prognosis and treatment response.

Our fully automated and reproducible CNV workflow includes extensive quality controls as well as meaningful graphical and tabular representations of CNV profiles. We perform analyses based on in-house generated data (e.g., low pass sequencing) as well as data from our panel sequencing, which can be customized to meet our clients' requirements.

Knowledge of the evolution of the tumor genome is of therapeutic interest, especially regarding prognosis, resistance formation and the development of targeted therapies. CNV-based tumor cell phylogenies (family trees) are an essential tool in this context. Since current molecular biology methods for DNA copy numbers are not exactly quantitative, both the relative and the absolute (actual) copy numbers must be reconstructed from the measured data.

We have developed a new statistically sound method that works even with low data quality. Our approach can be compared or supplemented with existing methods for complementary molecular data, such as mutations and so-called tandem repeats.

Based on our many years of experience in bioinformatics, data analysis and modeling, we offer you the development or adaptation of mathematical and bioinformatic methods as well as the implementation of software solutions within the framework of joint R&D projects. Our core competencies lie in the analysis of molecular data (so-called multi-omics data) in medicine, image analysis, data integration, biostatistics, machine learning and AI. We are also open to other fields of application like spatial omics, nanopore sequencing or big data sharing and future-oriented technologies such as quantum computing or federated learning. We also offer you the opportunity to work on these topics in collaboration with our Fraunhofer partner institutes, which specialize in these research areas.