SummaryThe Computational Biology Group (CBG) is active in the fields of computational biology, biostatistics, and systems biology. We develop, implement, and apply mathematical and statistical models for the analysis of high through-put molecular data to support the diagnosis and treatment of diseases. We are interested in disease-related perturbations of cellular networks and in the dynamics of evolutionary processes that lead to disease, such as the development of viral drug resistance and the somatic evolution of cancer.
Scientific progress in the biomedical sciences has largely been driven by technological innovations that allow for gathering large amounts of molecular data, such as DNA sequences, gene expression levels, or protein abundances. This molecular biology revolution has paved the way for quantitative biology. Mathematical, statistical, and computational methods are an integral part of systems biology today, but they also apply to clinical questions, giving rise to computational biomedicine. Our goal is to develop computational methods for making use of high-throughput molecular profiling data in a clinical setting. We employ probabilistic graphical models to describe cellular networks in healthy and diseased states, and we use evolutionary dynamics to model disease progression in infectious diseases and tumor progression in cancer.
Projects and Services
HIV therapy optimisation
Drug resistance is a major factor in therapy failure of HIV infected patients. The development of resistant mutants is an evolutionary process that is characterised by the accumulation of mutations. By integrating genomic and clinical data, we develop statistical models and efficient algorithms for analysing the HIV genome and for selecting optimal drug combinations. We have shown that the individual evolutionary potential of the HIV population within each single patient is predictive of therapy outcome. Our computational methods support diagnostics and personalised treatment of HIV infections.
Genetic diversity of pathogen populations
Genetic diversity is the consequence of random mutation and natural selection. Diverse and medically important pathogen populations include viral quasispecies, bacterial communities, and tumor cells. We use deep sequencing for inferring the genetic diversity of these mixed samples in a single sequencing run. We develop and apply mathematical and computational tools for correcting sequencing errors and for read assembly into a set of diverse haplotypes. Our initial approach makes use of Bayesian inference, statistical learning theory, and combinatorial optimisation. It is implemented in the software package ShoRAH.
Evolutionary dynamics of cancer
Cancer arises from alterations of the genome that transform cooperative cells in homeostatic tissues into non-cooperatively expanding mutants. This transformation typically requires multiple functional changes. Due to their fitness advantage, waves of clonally expanding cell lines are generated, each harboring additional mutations. We study the progression of cancer using evolutionary models and statistical inference based on whole-genome molecular profiling data.
Signaling networks of viral and bacterial infections
Many pathogens cannot be treated with present day drugs which target parasite gene products. An alternative therapeutic strategy is to interfere with host proteins that are essential for viral entry or bacterial uptake. We aim to understand the signaling cascades that are triggered in host cells upon infection. Our goal is to reconstruct, from RNA interference data, the cellular network responsible for pathogen sensing and inflammation and to identify host proteins as new potential drug targets. We develop statistical methods for the analysis of RNAi data and mathematical models for the reconstruction of signaling pathways employing probabilistic graphical models.
Website for Further Information
Computational Biology Group (CBG) www.cbg.ethz.ch