Biostatistics

We provide statistical consulting to ensure the right data and enough data are recorded to answer a scientific question. We help to condense difficult scientific questions into a clear, precise, specific mathematical language. We place particular emphasis on transparency and reproducibility in research.

We support clinical trials in statistical aspects at all stages from study conception to publication. In particular, we contribute to study protocols for Ethics committee submissions and grant applications, perform sample size determinations and power estimations, write statistical analysis plans, conduct data analyses, report trial results according to relevant guidelines, and assist with result interpretation and publication. We work in close collaboration with dTIP, the clinical trial platform at ETH.

We perform statistical analyses for any omics data (genomics, transcriptomics, proteomics, metagenomics, lipidomics, metabolomics, …) from any platform (micro-array, sequencing, mass-spectrometry, …). Multi-dimensional prediction from multiple omics data and modeling along distinct conditions or trajectories is our specialty.

We are using network analysis and genesets specific for diseases, pathways, drug mode of action, microbial identity, and more to lift raw omics results to a higher level. This facilitates a deeper understanding of disease-causing events, biological processes, structure activity relationships, or interaction with the microbiome.

We are specialized in the analysis of high-throughput screening data and can support your screening campaigns with selection of control conditions and Z'-factor calculation, data normalization and transformation, correction of plate spatial- or geometric patterns, and hit identification.

We provide statistical consulting and analysis of high-throughput compound screening campaigns and pharmacological or toxicological activity profiling. We support primarily the TDU, but we offer our services also to academic groups at ETH and elsewhere, and to customers from industry.

We are utilizing state-of-the-art methods of dimension reduction, denoising to decipher the complex, heterogenous and sometimes multi-modal data from single-cell experiments, for instance, (gene-, protein-, analyte-) marker discovery, celltype classification, trajectory reconstruction and differential subpopulation analysis.

We are experts in quantitative image analysis and in extracting predictive features relevant for a specific morphology and a clinical phenotype. We support translational and clinical research in the field of imaging cytometry, digital pathology, and medical imaging.

As we are routinely working with heterogeneous, complex, high-dimensional real-world data, our toolbox for predictive modeling ranges from well-interpretable linear models, like LASSO, more flexible classic machine learning, like random forests, to cutting edge deep learning algorithms, like convolutional neural networks.

Didn't find what you were looking for? us anyway! We implement new methods and topics regularly on the basis of new projects. Our experience with biomedical research and development puts us in an excellent position to efficiently acquire new knowhow and apply it in your projects.