Sven Bergmann (Link to main website)
 |
Sven Bergmann studied theoretical particle physics with Prof. Yosef Nir at the Weizmann Institute of Science (Israel) where he received his PhD in 2001. He then joined the laboratory of Prof. Naama Barkai in the Department of Molecular Genetics at the same institute, where he first worked as a Koshland postdoctoral fellow and later as staff scientist. He joined the Department of Medical Genetics in 2005 as Assistant Professor on a tenure track leading the Computational Biology Group and received tenure as Associate Professor in 2010. He is affiliated with the Swiss Institute of Bioinformatics since 2006. |
Research
Medical Genetics is undergoing an unprecedented revolution, and this revolution is happening at an ever increasing pace. There are two technological advances that are driving this revolution: First sequencing costs are dropping at an exponential rate, and as a consequence very soon complete genomes will be available for virtually everyone visiting a medical facility. On top of this we will also have high-throughput phenotypic data (metabolomics-, expression-, and epigenetic profiles) from accessible tissues in routine or dedicated tests. Diagnoses will increasingly depend on such molecular data, but this requires addressing immense challenges in terms of their analysis, due to their formidable size and potential noise. The only reason why there is hope in the endeavor is a parallel revolution.
This second revolution is maybe best characterized by Moore’s law, the fact that the computation power that can be bought for the same money roughly doubles approximately every two years. This exponential growth facilitates our present computational revolution that is felt in all aspects of life (internet, smart cell-phones, navigation, etc.) and this includes of course the means that we need to store and analyze the immense amount of biomedical data. This trend will continue and indeed it has to, if we are to have any chance in being able to come up with predictive algorithms for health risks and optimal treatment based on genomic data.
But mere computational power obviously is only a prerequisite, and we will need to develop novel approaches and algorithms to make the best use of the massive data. Our success will depend critically on the bringing together the people working at the bed- and bench-side. Only if computational researchers have a deep understanding of the biological and clinical challenges as well as limitations, can they develop the technologies to efficiently address them. Likewise, biologists and clinicians need to grasp the power as much as the constraints of computational analysis.
A focus of our research is to process the different types of data in order reduce their complexity. We do this by grouping fundamental elements (genes, samples, treatments) that share certain similarities into so-called "modules". This helps to make the huge amounts of data more amendable for analysis and may reveal the organization of and relation between the modules and their elements.
Such a modular approach is useful for each large dataset by itself: Analysis of massive gene expression data reveals "transcription modules" containing genes that are co-expressed in certain samples. Massive clinical data can be processed into "phenotypic modules" merging clinical variables that behave similarly across certain groups of individuals. Finally, we are exploring how higher-order structures in SNP data can be studied using "genotypic modules" containing SNP markers (or haplotypes) that co-segregate in subpopulations of a cohort.
The modular approach is also extendable for the joint analysis of different datasets. For example, we may identify a set of genes that is co-expressed in a group of cell-lines that all exhibit a similar sensitivity to a collection of treatments. Unsupervised identification of such "co-modules" provides testable hypotheses, e.g. we would predict that these genes are involved in pathways relevant for some cell-lines in response to related treatments.
A complementary direction of research pertains to relatively small genetic networks, whose components are well-known. We collaborate closely with experts of the field to identify biological systems that can be modeled quantitatively. Our goal in developing such models is not only to give an approximate description of system, but also to obtain a better understanding of its properties. For example, regulatory networks evolved to function reliably under ever-changing environmental conditions. This notion of robustness can guide computational analysis and provide constraints on models that complement those from direct measurements of the system's output.
In general, our group seeks an interdisciplinary approach, bridging the traditional gaps between physics, mathematics and biology. Our lab collaborates with experimental groups within and outside our department. In particular, due to our proximity to the CHUV we have close contacts to medical research groups and assist the analysis of clinical data.
Selected Publications (click here for full list)
Fisun Hamaratoglu*, Aitana Morton de Lachapelle*, George Pyrowolakis, Sven Bergmann*, Markus Affolter* (*: equal authors) Dpp Signaling Activity Requires Pentagone to Scale with Tissue Size in the Growing Drosophila Wing Imaginal Disc, PLoS Biology (article), in press.
Kutalik, Z. et al. Methods for testing association between uncertain genotypes and quantitative traits. Biostatistics 12, 1-17 (2011).
Kutalik, Z. et al. Novel Method to Estimate the Phenotypic Variation Explained by Genome-Wide Association Studies Reveals Large Fraction of the Missing Heritability. Genetic Epidemiology 35, 341-349 (2011).
Schupbach, T., Xenarios, I., Bergmann, S. & Kapur, K. FastEpistasis: a high performance computing solution for quantitative trait epistasis. Bioinformatics 26, 1468-9 (2010).
Henrichsen, C.N. et al. Using transcription modules to identify expression clusters perturbed in Williams-Beuren syndrome. PLoS Comput Biol 7, e1001054 (2011).
de Lachapelle, A.M. & Bergmann, S. Pre-steady and stable morphogen gradients: can they coexist? Mol Syst Biol 6 (2010).
de Lachapelle, A.M. & Bergmann, S. Precision and scaling in morphogen gradient read-out. Mol Syst Biol 6, 351 (2010).
Csardi, G., Kutalik, Z. & Bergmann, S. Modular analysis of gene expression data with R. Bioinformatics 26, 1376-7 (2010).
Luscher, A. et al. ExpressionView-an interactive viewer for modules identified in gene expression data. Bioinformatics 26, 2062-2063 (2010).
Kapur, K. et al. Genome-Wide Meta-Analysis for Serum Calcium Identifies Significantly Associated SNPs near the Calcium-Sensing Receptor (CASR) Gene. Plos Genetics 6 (2010).
Rauch, A. et al. Genetic Variation in IL28B Is Associated With Chronic Hepatitis C and Treatment Failure: A Genome-Wide Association Study. Gastroenterology 138, 1338-U173 (2010).
Hor, H. et al. Genome-wide association study identifies new HLA class II haplotypes strongly protective against narcolepsy. Nat Genet 42, 786-9 (2010).
Novembre, J. et al. Genes mirror geography within Europe. Nature 456, 98-101 (2008).
Kutalik Z, Beckmann JS and Bergmann S (2008). A modular approach for integrative analysis of large-scale gene-expression and drug-response data Nature Biotechnology 26(5): 531-9.
