Band 5, Heft 2 - Special Issue: 5th International Symposium on Integrative Bioinformatics, Lutherstadt Wittenberg, Germany, 2008 / Guest Editors: Matthias Lange, Uwe Scholz, Falk Schreiber, Ralf Hofestädt, Jacob Köhler, Paul Verrier

Background: A precise experimental identification of transcription factor binding motifs (TFBMs), accurate to a single base pair, is time-consuming and difficult. For several databases, TFBM annotations are extracted from the literature and stored 5ʹ → 3ʹ relative to the target gene. Mixing the two possible orientations of a motif results in poor information content of subsequently computed position frequency matrices (PFMs) and sequence logos. Since these PFMs are used to predict further TFBMs, we address the question if the TFBMs underlying a PFM can be re-annotated automatically to improve both the information content of the PFM and subsequent classification performance. Results: We present MoRAine, an algorithm that re-annotates transcription factor binding motifs. Each motif with experimental evidence underlying a PFM is compared against each other such motif. The goal is to re-annotate TFBMs by possibly switching their strands and shifting them a few positions in order to maximize the information content of the resulting adjusted PFM. We present two heuristic strategies to perform this optimization and subsequently show that MoRAine significantly improves the corresponding sequence logos. Furthermore, we justify the method by evaluating specificity, sensitivity, true positive, and false positive rates of PFM-based TFBM predictions for E. coli using the original database motifs and the MoRAine-adjusted motifs. The classification performance is considerably increased if MoRAine is used as a preprocessing step. Conclusions: MoRAine is integrated into a publicly available web server and can be used online or downloaded as a stand-alone version from http://moraine.cebitec.uni-bielefeld.de.

The Gene Expression Management System (GEMS) is a database system for patterns of gene expression. These patterns result from systematic whole-mount fluorescent in situ hybridization studies on zebrafish embryos. GEMS is an integrative platform that addresses one of the important challenges of developmental biology: how to integrate genetic data that underpin morphological changes during embryogenesis. Our motivation to build this system was by the need to be able to organize and compare multiple patterns of gene expression at tissue level. Integration with other developmental and biomolecular databases will further support our understanding of development. The GEMS operates in concert with a database containing a digital atlas of zebrafish embryo; this digital atlas of zebrafish development has been conceived prior to the expansion of the GEMS. The atlas contains 3D volume models of canonical stages of zebrafish development in which in each volume model element is annotated with an anatomical term. These terms are extracted from a formal anatomical ontology, i.e. the Developmental Anatomy Ontology of Zebrafish (DAOZ). In the GEMS, anatomical terms from this ontology together with terms from the Gene Ontology (GO) are also used to annotate patterns of gene expression and in this manner providing mechanisms for integration and retrieval . The annotations are the glue for integration of patterns of gene expression in GEMS as well as in other biomolecular databases. At the one hand, zebrafish anatomy terminology allows gene expression data within GEMS to be integrated with phenotypical data in the 3D atlas of zebrafish development. At the other hand, GO terms extend GEMS expression patterns integration to a wide range of bioinformatics resources.

This paper presents a novel bioinformatics data warehouse software kit that integrates biological information from multiple public life science data sources into a local database management system. It stands out from other approaches by providing up-to-date integrated knowledge, platform and database independence as well as high usability and customization. This open source software can be used as a general infrastructure for integrative bioinformatics research and development. The advantages of the approach are realized by using a Java-based system architecture and object-relational mapping (ORM) technology. Finally, a practical application of the system is presented within the emerging area of medical bioinformatics to show the usefulness of the approach. The BioDWH data warehouse software is available for the scientific community at http://sourceforge.net/projects/biodwh/.

The automated annotation of data from high throughput sequencing and genomics experiments is a significant challenge for bioinformatics. Most current approaches rely on sequential pipelines of gene finding and gene function prediction methods that annotate a gene with information from different reference data sources. Each function prediction method contributes evidence supporting a functional assignment. Such approaches generally ignore the links between the information in the reference datasets. These links, however, are valuable for assessing the plausibility of a function assignment and can be used to evaluate the confidence in a prediction. We are working towards a novel annotation system that uses the network of information supporting the function assignment to enrich the annotation process for use by expert curators and predicting the function of previously unannotated genes. In this paper we describe our success in the first stages of this development. We present the data integration steps that are needed to create the core database of integrated reference databases (UniProt, PFAM, PDB, GO and the pathway database Ara- Cyc) which has been established in the ONDEX data integration system. We also present a comparison between different methods for integration of GO terms as part of the function assignment pipeline and discuss the consequences of this analysis for improving the accuracy of gene function annotation. The methods and algorithms presented in this publication are an integral part of the ONDEX system which is freely available from http://ondex.sf.net/.

Biomolecular modelling has provided computational simulation based methods for investigating biological processes from quantum chemical to cellular levels. Modelling such microscopic processes requires atomic description of a biological system and conducts in fine timesteps. Consequently the simulations are extremely computationally demanding. To tackle this limitation, different biomolecular models have to be integrated in order to achieve high-performance simulations. The integration of diverse biomolecular models needs to convert molecular data between different data representations of different models. This data conversion is often non-trivial, requires extensive human input and is inevitably error prone. In this paper we present an automated data conversion method for biomolecular simulations between molecular dynamics and quantum mechanics/molecular mechanics models. Our approach is developed around an XML data representation called BioSimML (Biomolecular Simulation Markup Language). BioSimML provides a domain specific data representation for biomolecular modelling which can effciently support data interoperability between different biomolecular simulation models and data formats.

A theoretical basis for the evaluation of the effciency of quarantine measure is developed in a SIR model with time delay. In this model, the effectiveness of the closure of public places such as schools in disease control, modeled as a high degree node in a social network, is evaluated by considering the effect of the time delay in the identification of the infected. In the context of the SIR model, the relation between the number of infectious individuals who are identified with time delay and then quarantined and those who are not identified and continue spreading the virus are investigated numerically. The social network for the simulation is modeled by a scale free network. Closure measures are applied to those infected nodes with high degrees. The effectiveness of the measure can be controlled by the present value of the critical degree K C : only those nodes with degree higher than K C will be quarantined. The cost C Q incurred for the closure measure is assumed to be proportional to the total links rendered inactive as a result of the measure, and generally decreases with KC, while the medical cost C Q incurred for virus spreading increases with K C . The total social cost (C M + C Q ) will have a minimum at a critical K C , which depends on the ratio of medical cost coeffcient α M and closure cost coeffcient α Q . Our simulation results demonstrate a mathematical procedure to evaluate the effciency of quarantine measure. Although the numerical work is based on a scale free network, the procedure can be readily generalized and applied to a more realistic social network to determine the proper closure measure in future epidemics.

The need for novel methods of visualizing microarray data is growing. New perspectives are beneficial to finding patterns in expression data. The Bluejay genome browser provides an integrative way of visualizing gene expression datasets in a genomic context. We have now developed the functionality to display multiple microarray datasets simultaneously in Bluejay, in order to provide researchers with a comprehensive view of their datasets linked to a graphical representation of gene function. This will enable biologists to obtain valuable insights on expression patterns, by allowing them to analyze the expression values in relation to the gene locations as well as to compare expression profiles of related genomes or of di erent experiments for the same genome.

We have developed a computational pipeline to analyse large surveys of A ymetrix GeneChips, for example NCBI’s Gene Expression Omnibus. GEO samples data for many organisms, tissues and phenotypes. Because of this experimental diversity, any observed correlations between probe intensities can be associated either with biology that is robust, such as common co-expression, or with systematic biases associated with the GeneChip technology. Our bioinformatics pipeline integrates the mapping of probes to exons, quality control checks on each GeneChip which identifies flaws in hybridization quality, and the mining of correlations in intensities between groups of probes. The output from our pipeline has enabled us to identify systematic biases in GeneChip data. We are also able to use the pipeline as a discovery tool for biology. We have discovered that in the majority of cases, A ymetrix probesets on Human GeneChips do not measure one unique block of transcription. Instead we see numerous examples of outlier probes. Our study has also identified that in a number of probesets the mismatch probes are an informative diagnostic of expression, rather than providing a measure of background contamination. We report evidence for systematic biases in GeneChip technology associated with probe-probe interactions. We also see signatures associated with post-transcriptional processing of RNA, such as alternative polyadenylation.

We present a Bayes-Random Fields framework which is capable of integrating unlimited data sources for discovering relevant network architecture of large-scale networks. The random field potential function is designed to impose a cluster constraint, teamed with a full Bayesian approach for incorporating heterogenous data sets. The probabilistic nature of our framework facilitates robust analysis in order to minimize the influence of noise inherent in the data on the inferred structure in a seamless and coherent manner. This is later proved in its applications to both large-scale synthetic data sets and Saccharomyces Cerevisiae data sets. The analytical and experimental results reveal the varied characteristic of different types of data and reflect their discriminative ability in terms of identifying direct gene interactions.

We describe a method for querying vertex- and edge-labeled graphs using context-free grammars to specify the class of interesting paths. We introduce a novel problem, finding the connection subgraph induced by the set of matching paths between given two vertices or two sets of vertices. Such a subgraph provides a concise summary of the relationship between the vertices. We also present novel algorithms for parsing subgraphs directly without enumerating all the individual paths. We evaluate experimentally the presented parsing algorithms on a set of real graphs derived from publicly available biomedical databases and on randomly generated graphs. The results indicate that parsing the connection subgraph directly is much more effective than parsing individual paths separately. Furthermore, we show that using a bidirectional parsing algorithm, in most cases, allows for searching twice as long paths as using a unidirectional search strategy.

Proteins and their interactions are essential for the functioning of all organisms and for understanding biological processes. Alternative splicing is an important molecular mechanism for increasing the protein diversity in eukaryotic cells. Splicing events that alter the protein structure and the domain composition can be responsible for the regulation of protein interactions and the functional diversity of different tissues. Discovering the occurrence of splicing events and studying protein isoforms have become feasible using Affymetrix Exon Arrays. Therefore, we have developed the versatile Cytoscape plugin DomainGraph that allows for the visual analysis of protein domain interaction networks and their integration with exon expression data. Protein domains affected by alternative splicing are highlighted and splicing patterns can be compared.

Summar y ReMatch is a web-based, user-friendly tool that constructs stoichiometric network models for metabolic flux analysis, integrating user-developed models into a database collected from several comprehensive metabolic data resources, including KEGG, MetaCyc and CheBI. Particularly, ReMatch augments the metabolic reactions of the model with carbon mappings to facilitate 13 C metabolic flux analysis. The construction of a network model consisting of biochemical reactions is the first step in most metabolic modelling tasks. This model construction can be a tedious task as the required information is usually scattered to many separate databases whose interoperability is suboptimal, due to the heterogeneous naming conventions of metabolites in different databases. Another, particularly severe data integration problem is faced in 13 C metabolic flux analysis, where the mappings of carbon atoms from substrates into products in the model are required. ReMatch has been developed to solve the above data integration problems. First, ReMatch matches the imported user-developed model against the internal ReMatch database while considering a comprehensive metabolite name thesaurus. This, together with wild card support, allows the user to specify the model quickly without having to look the names up manually. Second, ReMatch is able to augment reactions of the model with carbon mappings, obtained either from the internal database or given by the user with an easy-touse tool. The constructed models can be exported into 13C-FLUX and SBML file formats. Further, a stoichiometric matrix and visualizations of the network model can be generated. The constructed models of metabolic networks can be optionally made available to the other users of ReMatch. Thus, ReMatch provides a common repository for metabolic network models with carbon mappings for the needs of metabolic flux analysis community. ReMatch is freely available for academic use at http://www.cs.helsinki.fi/group/sysfys/software/rematch/.

A new look at the transcription start is presented in which we can see transcription factors binding to both sides of the TSS as an essential requirement. Naturally the factor binding to the downstream region must be removed so that transcription process can continue. The presence of a number of distinct transcription factors also can be used to explain selective activation of various genes. The transcription start site by itself plays only a minor role in the whole process. We also suggest that mutations close to the TSS on the coding side can be fatal even if preserves the codon table.

The functional annotation of genomic data has become a major task for the ever-growing number of sequencing projects. In order to address this challenge, we recently developed GOblet, a free web service for the annotation of anonymous sequences with Gene Ontology (GO) terms. However, to overcome limitations of the GO terminology, and to aid in understanding not only single components but as well systemic interactions between the individual components, we have now extended the GOblet web service to integrate also pathway annotations. Furthermore, we extended and upgraded the data analysis pipeline with improved summaries, and added term enrichment and clustering algorithms. Finally, we are now making GOblet available as a stand-alone application for high-throughput processing on local machines. The advantages of this frequently requested feature is that a) the user can avoid restrictions of our web service for uploading and processing large amounts of data, and that b) confidential data can be analysed without insecure transfer to a public web server. The stand-alone version of the web service has been implemented using platform independent Tcl-scripts, which can be run with just a single runtime file utilizing the Starkit technology. The GOblet web service and the stand-alone application are freely available at http://goblet.molgen.mpg.de.

Traditional analysis of gene expression profiles use clustering to find groups of coexpressed genes which have similar expression patterns. However clustering is time consuming and could be diffcult for very large scale dataset. We proposed the idea of Discovering Distinct Patterns (DDP) in gene expression profiles. Since patterns showing by the gene expressions reveal their regulate mechanisms. It is significant to find all different patterns existing in the dataset when there is little prior knowledge. It is also a helpful start before taking on further analysis. We propose an algorithm for DDP by iteratively picking out pairs of gene expression patterns which have the largest dissimilarities. This method can also be used as preprocessing to initialize centers for clustering methods, like K-means. Experiments on both synthetic dataset and real gene expression datasets show our method is very effective in finding distinct patterns which have gene functional significance and is also effcient.

With a large amount of information relating to proteins accumulating in databases widely available online, it is of interest to apply machine learning techniques that, by extracting underlying statistical regularities in the data, make predictions about the functional and evolutionary characteristics of unseen proteins. Such predictions can help in achieving a reduction in the space over which experiment designers need to search in order to improve our understanding of the biochemical properties. Previously it has been suggested that an integration of features computable by comparing a pair of proteins can be achieved by an artificial neural network, hence predicting the degree to which they may be evolutionary related and homologous. We compiled two datasets of pairs of proteins, each pair being characterised by seven distinct features. We performed an exhaustive search through all possible combinations of features, for the problem of separating remote homologous from analogous pairs, we note that significant performance gain was obtained by the inclusion of sequence and structure information. We find that the use of a linear classifier was enough to discriminate a protein pair at the family level. However, at the superfamily level, to detect remote homologous pairs was a relatively harder problem. We find that the use of nonlinear classifiers achieve significantly higher accuracies. In this paper, we compare three different pattern classification methods on two problems formulated as detecting evolutionary and functional relationships between pairs of proteins, and from extensive cross validation and feature selection based studies quantify the average limits and uncertainties with which such predictions may be made. Feature selection points to a “knowledge gap” in currently available functional annotations. We demonstrate how the scheme may be employed in a framework to associate an individual protein with an existing family of evolutionarily related proteins.