- KonferenzbeitragRegistration to a neuroanatomical reference atlas - identifying glomeruli in optical recordings of the honeybee brain(German Conference on Bioinformatics, 2008) Strauch, Martin; Galizia, C. Giovanni; Beyer, Andreas; Schroeder, MichaelAn odorant stimulus given to a bee elicits a characteristic combinatorial pattern of activity in neuronal units called glomeruli. These patterns can be measured by optical imaging, however detecting and identifying the glomeruli is a laborious task and prone to errors. Here, we present an image analysis pipeline for the automatic detection and identification of glomeruli. It involves Independent Component Analysis (ICA) to detect glomeruli in CCD camera data, a filtering step to exclude non- glomerulus objects and a graph-matching approach to find the best projection of the observed brain region onto a reference atlas. We evaluate our method against a manual glomerulus identification performed by a human expert and show that we achieve reliable results. Employing our method, we are now able to screen multiple recordings with the same accuracy, yielding a homogeneous collection of glomerulus identity mappings. These will subsequently be used to extract activity patterns that can be compared between individuals.
- KonferenzbeitragProtein Structure Alignment through a Contact Topology Profile using SABERTOOTH(German Conference on Bioinformatics, 2008) Teichert, F.; Bastolla, U.; Porto, M.; Beyer, Andreas; Schroeder, MichaelThe contact vector (CV) of a protein structure is one of the simplest and most condensed descriptions of protein structure available. It lists the number of con- tacts each amino acid has with the surrounding structure and has frequently been used e.g. to derive approximative folding energies in protein folding analysis. The CV, however, is a lossy structure representation, as it does not contain sufficient information to allow for the reconstruction of the full protein structure it was derived from. The loss of information leads to a degeneracy in the sense that a single contact vector is compatible with many different contact matrices, but it has been shown that this degeneracy is nearly fully compensated by the physical constraints protein structure is subject to. We recently developed the alignment framework ‘SABERTOOTH’ that is able to generically align connectivity related vectorial structure profiles to compute protein alignments. Here we show that also the CV allows for state-of-the-art alignment quality, just like the elaborated ‘Effective Connectivity’ profile (EC) that SABERTOOTH currently uses. This simplification leeds to a very simple and elegant approach to structure alignment, which accelerates and generalizes the algorithm we previously proposed. Furthermore, we conclude from our work that the CV in itself is a useful structure description if its collective properties are called for.
- KonferenzbeitragDesigning Binding Pockets on Protein Surfaces using the A* Algorithm(German Conference on Bioinformatics, 2008) Eyrisch, Susanne; Helms, Volkhard; Beyer, Andreas; Schroeder, MichaelThe in-silico design of ligands binding to the protein surface instead of deep binding pockets is still a great challenge. Often no appropriate binding pockets are available in the apo experimental structures and standard virtual screening techniques will fail. Here, we present two new algorithms for designing tailored ligand binding pockets on the protein surface that account for protein backbone and side chain flexibility. At first, the protein surface is scanned for potential pocket positions using a program named PocketScanner. This program minimizes the protein energetically in the presence of generic pocket spheres representing the new binding pockets whose positions remain fixed. The side chains of the relaxed protein conformations are then further refined by a second program named PocketBuilder. PocketBuilder identifies all residues within a given radius of the pocket positions and searches for the best combination of side chain rotamers using the A* algorithm. Given multiple protein conformations as input, PocketBuilder identifies those that lead to the best results, namely protein conformations of low energy that possess binding pockets with desired properties. The approach was tested on the proteins BCL-XL, IL-2, and MDM2 which are involved in protein-protein interactions and hence represent challenging drug tar- gets. Although the native ligand binding pocket was not or only partly open in the apo crystal or NMR structures, PocketScanner and PocketBuilder successfully generated conformations with pockets into which a known inhibitor could be docked in a native- like orientation for two out of the three test systems. For BCL-XL, the docking scores were even similar to those obtained in re-docking experiments to the inhibitor bound crystal structure.
- KonferenzbeitragA Propagation-based Algorithm for Inferring Gene-Disease Associations(German Conference on Bioinformatics, 2008) Vanunu, Oron; Sharan, Roded; Beyer, Andreas; Schroeder, MichaelA fundamental challenge in human health is the identification of disease- causing genes. Recently, several studies have tackled this challenge via a two-step approach: first, a linkage interval is inferred from population studies; second, a computational approach is used to prioritize genes within this interval. State-of-the-art methods for the latter task are based on the observation that genes causing the same or similar diseases tend to lie close to one another in a network of protein-protein or functional interactions. However, most of these approaches use only local network information in the inference process. Here we provide a global, network-based method for prioritizing disease genes. The method is based on formulating constraints on the prioritization function that relate to its smoothness over the network and usage of prior information. A propagation-based method is used to compute a function satisfying the constraints. We test our method on gene-disease association data in a cross-validation setting, and compare it to extant prioritization approaches. We show that our method provides the best overall performance, ranking the true causal gene first for 29% of the 1,369 diseases with a known gene in the OMIM knowledgebase.
- KonferenzbeitragEvolutionary Construction of Multiple Graph Alignments for the Structural Analysis of Biomolecules(German Conference on Bioinformatics, 2008) Fober, Thomas; Hüllermeier, Eyke; Mernberger, Marco; Beyer, Andreas; Schroeder, MichaelThe concept of multiple graph alignment has recently been introduced as a novel method for the structural analysis of biomolecules. Using inexact, approximate graph-matching techniques, this method enables the robust identification of approximately conserved patterns in biologically related structures. In particular, multiple graph alignments enable the characterization of functional protein families independent of sequence or fold homology. This paper first recalls the concept of multiple graph alignment and then addresses the problem of computing optimal alignments from an algorithmic point of view. In this regard, a method from the field of evolutionary algorithms is proposed and empirically compared to a hitherto existing greedy strategy. Empirically, it is shown that the former yields significantly better results than the latter, albeit at the cost of an increased runtime.
- KonferenzbeitragIdentifying the topology of protein complexes from affinity purification assays(German Conference on Bioinformatics, 2008) Friedel, Caroline C.; Zimmer, Ralf; Beyer, Andreas; Schroeder, MichaelRecent advances in high-throughput technologies have made it possible to investigate not only individual protein interactions but the association of these proteins in complexes. So far the focus has been on the prediction of complexes as sets of proteins from the experimental results while the modular substructure and the physical interactions within protein complexes have been mostly ignored. In this article, we present an approach for identifying the direct physical interactions and the subcomponent structure of protein complexes predicted from affinity purification assays. Our algorithm calculates the union of all maximum spanning trees from scoring networks for each protein complex to extract relevant interactions. In a subsequent step this network is extended to interactions which are not accounted for by alternative indirect paths. We show that the interactions identified with this approach are more accurate in predicting experimentally derived physical interactions than baseline approaches and resolve more satisfactorily the subcomponent structure of the complexes. The usefulness of our approach is illustrated on the RNA polymerases for which the modular substructure can be successfully reconstructed with our method.
- KonferenzbeitragExploring the Enzyme Neighbourhood to interpret gene expression data(German Conference on Bioinformatics, 2008) Goffard, Nicolas; Frickey, Tancred; Imin, Nijat; Weiller, Georg; Beyer, Andreas; Schroeder, MichaelPost-genomic data analysis represents a new challenge to link and interpret the vast amount of raw data obtained with transcriptomic or proteomic techniques in the context of metabolic pathways. We propose a new strategy with the help of a metabolic network graph to extend PathExpress, a web-based tool to interpret gene expression data, without being restricted to predefined pathways. We defined the Enzyme Neighbourhood as groups of linked enzymes, corresponding to a sub-network, to explore the metabolic network in order to identify the most relevant sub-networks affected in gene expression experiments.
- KonferenzbeitragResqMi - a versatile algorithm and software for Resequencing Microarrays(German Conference on Bioinformatics, 2008) Symons, Stephan; Weber, Kirstin; Bonin, Michael; Nieselt, Kay; Beyer, Andreas; Schroeder, MichaelResequencing microarrays are a common tool for fast monitoring of individual genetic variations. Applications include diagnosis of genetic and infectious diseases and SNP prediction. Base calling is the crucial step in the analysis of resequencing data. All current base calling algorithms produce ambiguous calls on parts of the sequence. Therefore, proper data handling, editing and visualization as well as revised calling algorithms are generally necessary for successful data interpretation. We present a base calling algorithm that uses a model-based approach using intensity comparisons and region-wise conformance assessment, as well as an algorithm to revise uncalled positions. The calling algorithm is shown to have call rates comparable to ABACUS, the currently most commonly used method. Both algorithms combined however can considerably increase the calling rate. We also present a new open source software called ResqMi, short for Resequencing using Microarrays, which focuses on the efficient and user-friendly analysis, visual inspection and easy manual editing of resequencing microarray data. Both algorithms are implemented as plugins for ResqMi. ResqMi is available at http://www-ps.informatik.uni-tuebingen.de/resqmi
- Editiertes BuchGerman Conference on Bioinformatics(German Conference on Bioinformatics, 2008) Beyer, Andreas; Schroeder, Michael
- KonferenzbeitragFRANz: Fast reconstruction of wild pedigrees(German Conference on Bioinformatics, 2008) Riester, Markus; Peter F., Stadler; Klemm, Konstantin; Beyer, Andreas; Schroeder, MichaelWe present a software package for fast pedigree reconstruction in natural populations using co-dominant genomic markers such as microsatellites and SNPs. If available, the algorithm makes use of prior information such as known relationships (sub-pedigrees) or the age and sex of individuals. Statistical confidence is estimated by a simulation of the sampling process. The accuracy of the algorithm is demonstrated for simulated data as well as an empirical data set with known pedigree. The parentage inference is robust even in the presence of genotyping errors.