BIOTEC - Biotechnology Center TU Dresden

A growing number of technologies allow for the genome-scale measurement of biological properties such as protein and mRNA concentrations or phenotypic changes (e.g. response to RNAi knock-downs). The genome-wide nature of the available data facilitates a systems perspective: It becomes possible to go beyond individual genes or pathways and to study regulatory processes of the entire system ‘cell’. However, up to now the potential is by far not being fully exploited.

Our group adopts a network perspective by studying relationships between proteins and other biomolecules (e.g. DNA, RNA) in silico to reveal the regulatory context of relevant genes. During the past years we contributed new computational methods for large-scale data integration, network biology, and statistical genetics. Even though we do not do any experiments ourselves, we have a tight network of experimental collaborators and together with them we develop experiments that support our computational analysis.

Network Reconstruction
First, regulatory networks have to be uncovered, which we achieve by integrating a wide range of different data sets originating from public databases and from our collaborators. One important step in gene expression regulation is the interaction between transcription factors and their target genes. We presented the most comprehensive method for combining all possible experimental and computational evidences indicating an interaction between a transcription factor and a potential target gene (Beyer et al., 2006). In addition to identifying those interactions between transcription factors and their targets we also developed an algorithm to disclose the complex interactions among transcription factors (“combinatorial regulation”).

The experimental detection of protein-protein interactions is an important contribution to modern systems biology. Even though advanced technologies are used, it remains impossible to completely reveal the human ‘interactome’ experimentally. Thus, we also work on developing tools for the computational prediction of physical protein-protein interactions. Our interaction networks are subsequently used for guiding experimental efforts and for the integrated analysis with other genomic data.

Post-transcriptional Regulation
Unlike many others we study gene expression regulation also at the post-transcriptional level. Previously, we demonstrated the importance of post-transcriptional regulation and we are pioneering new ways of analysing those processes at genomic scale (Beyer et al., 2004). We were able to link protein functions to specific regulatory patterns, such as ‘preferential transcriptional regulation’ or ‘preferentially regulated via protein turnover’, etc. Furthermore, we coined the term ‘translation on demand’, which refers to a mechanism by which cells can quickly increase the synthesis of specific proteins under stress (Beyer et al., 2004, Brockmann et al., 2007).

Analysis of high-dimensional RNAi screens
An increasing number of RNAi screens is characterizing the knock-down phenotypes by many parameters. High-throughput technologies such as automated image analysis or FACS allow for the simultaneous measurement of several parameters for every single knock-down. However, the computational analysis of the resulting data is challenging. The noise in the data is posing unprecedented problems in such high-dimensional screens. Further, the biological interpretation of the data is often elusive. We help by providing new methods that map RNAi screening data onto interaction network, which (1) removes noise especially due to off-target-effects, and (2) aids the identification of molecular pathways that mediate the observed phenotypes.

Explaining the impact of natural genetic variability on physiological phenotypes
Natural genetic variation is determining someone’s eye and hair colour. Yet, other traits such as disease susceptibility are also affected by genetic variations. In order to improve our understanding of complex diseases and to support the development of new diagnostic methods and treatments, we develop systems biology methods for linking genetic variation to phenotypic variation. Our ultimate goal is to understand the molecular mechanisms linking the two together.

For example in the EU funded project PhenOxiGEn we aim to understand how genetic variation in a collection of yeast strains causes changes in a cell’s ability to respond to oxidative stress. In this collaborative project we perform a range of large-scale experiments and integrate the data with new pathway modelling techniques.

In another project we are analysing expression QTL (eQTL) data from our collaborators Gerd Kempermann (CRTD) an Andrew Su (GNF, California) in order to reveal the pathways controlling adult neurogenesis in the hippocampus. Here we are applying network-based analysis methods to clean up the eQTL data and to mechanistically explain the causal relationships between significant loci and the respective target genes.

In the future we will specifically design experiments with our collaborators that will be perfectly tailored for our models. These data will be integrated at a yet higher level in order to uncover the tight linking between transcriptional and post-transcriptional regulatory pathways in model species and human cell lines. This will address questions such as:

• How are different stress response pathways or developmental pathways interlinked?
• Many pathways control expression at different levels (transcription, RNA-turnover, translation, etc.). Where are the 'branching points' of such pathways?
• How do small regulatory RNAs and transcription factors interact to control gene expression?

This network visualizes the complex hierarchical organization of transcriptional regulation in Saccharomyces cerevisiae. Each node represents a distinct set of transcription factors, where downstream modules (at bottom) are composed as combinations of upstream modules (at top).

Our group is part of the DFG-Center for Regenerative Therapies Dresden (CRTD) and we are a member of the Dresden International Graduate School for Biomedicine and Bioengineering.

Michaelson JJ, Loguercio S, Beyer A. (2009) Detection and interpretation of expression quantitative trait loci (eQTL). Methods 48(3):265-76.

Suthram S, Beyer A, Ideker T. (2008) eQED: an efficient method for interpreting eQTL associations using protein networks. Molec. Syst. Biol. 4:162.

Beyer A, Bandyopadhyay S, Ideker T. (2007) Integrating physical and genetic maps: from genomes to interaction networks. Nat. Rev. Genet. 8(9):699-710.

R. Brockman, A.Beyer, J. Heinisch, T. Wilhelm (2007) Posttranscriptional expression regulation: what determines translation rates? PLoS Comput. Biol. 3(3):e57.

A. Beyer, C. Workman, J. Hollunder, D. Radke, U. Möller, T. Wilhelm, T.G. Ideker (2006) Integrated assessment and prediction of transcription factor binding. PLoS Comput. Biol. 2(6):e70.

J. Hollunder, A. Beyer, T. Wilhelm (2005) Identification and characterization of protein subcomplexes in yeast. Proteomics 5(8):2082-9.

A. Beyer, T. Wilhelm (2005) Dynamic simulation of protein complex formation on a genomic scale. Bioinformatics 21(8):1610-6.

A. Beyer, J. Hollunder, H.P. Nasheuer, T. Wilhelm (2004) Post-transcriptional expression regulation in the yeast Saccharomyces cerevisiae on a genomic scale. Mol. Cell. Proteomics 3(11):1083-92.

Find more publications on PubMed.
ResearcherID

• 

  2002 PhD (Systems Science) at the University of Osnabrück, Germany
• 2002 – 2006: Postdoctoral work at the Leibniz Institute for Age Research, Jena (Thomas Wilhelm) and the University of California San Diego (Trey Ideker)
• Since 2007: Group leader “Cellular Networks & Systems Biology” at the BIOTEC, TU-Dresden.

Want to join the team? This is your chance!

Becoming a PhD Student
Our group is a member of the Dresden International Graduate School for Biomedicine and Bioengineering. Students in this prestigious program get many additional benefits, such as special training, German language classes and support of a strong community of other peers in Dresden. Please, apply through this program to become a PhD student in my group. Informal inquiries are welcome.

All current group members are listed on the Staff Page.

Here you find data and programming code made available by the Beyer-Group.

Human Predicted Protein-Protein Interactions (hPRINT)

A predicted genome-wide human interactome. We integrated a wide range of computational evidences for the de-novo prediction of physical protein interactions. The database also lists known, experimentally tested interactions from various other popular sources. (coming soon)

Random Forest QTL Mapping

Mapping quantitaive traits using Random Forest. This method accounts for possible non-additive (epistatic) interactions between loci. (coming soon)