Information Technology and Systems - 2011
Conference for Young Scientists and Engineers
October 2 – 7, 2011

| English





Monday, October 3
14:30 - 16:30
Ballroom A
Session: Bioinformatics - RECESS (eng)
Chair: Dr. Mikhail Gelfand

Ludwig Geistlinger, Gergely Csaba, Robert Küffner, Nicola Mulder, Ralf Zimmer
From Sets To Graphs: Towards a Realistic Enrichment Analysis of Transcriptomic Systems Downoad paper
Abstract: Current gene set enrichment approaches do not take interactions and associations between set members into account. We analyzed established gene set enrichment methods and their result sets in a large-scale investigation of 1000 expression datasets. The reported statistically significant gene sets exhibit only average consistency between the observed patterns of differential expression and known regulatory interactions. We present Gene Graph Enrichment Analysis (GGEA) to detect consistently and coherently enriched gene sets, based on prior knowledge derived from directed gene regulatory networks (GRNs). GGEA significantly increases the detection of gene sets where measured positively or negatively correlated expression patterns coincide with directed inducing or repressing relationships thus facilitating further interpretation of gene expression data.

Constanze Schmitt, Matthias Boeck, Stefan Kramer
SOM Biclustering of Gene Expression Data Downoad paper
Abstract: Self-Organising Maps (SOMs) are an unsupervised learning mechanism mainly used for dimensionality reduction of high-dimensional data. This makes them particularly useful when dealing with gene expression microarray data, where they are invaluable for exploratory data analysis, such as cluster identification. The classical SOM approach performs clustering in only one dimension. However, with multiple gene expression chips describing different experimental conditions or individuals, biclustering is far more suitable to detect patterns of co-expressed genes present in only a subset of the samples. Therefore, biclustering variants of SOMs would be required to transfer the advantages of SOMs to the world of gene expression bicluster analysis. This paper describes SOM-based biclustering approaches, in particular the approach by Cottrel et al. (Korresp) and one proposed extension, and assesses them on synthetic and biological data.

Matthias Boeck, Constanze Schmitt, Stefan Kramer
A Study of Dynamic Time Warping for the Inference of Gene Regulatory Relationships Downoad paper
Abstract: In this study we assess different variants of Dynamic Time Warping (DTW) for the inference of gene regulatory relationships. Apart from DTW on continuous time series, we present a novel angle-based discretization approach and a distance learning method that is combined with DTW to find new gene interactions. A positive influence of the distance optimization on the performance of the alignments of gene expression profiles could not yet be established. However, our results show that discretization can be important to the outcome of the alignments. The discretization is not only able to keep the important features of the time series, it is also able to perform better than regular DTW on the original data.

Robert Pesch
To transfer or not to transfer - Complementing the eukaryotic protein-protein interactome Downoad paper
Abstract: Many approaches have been presented to automate the annotation of proteins in various species, but still only a fraction of proteins have a detailed functional description. Such functional descriptions can be used to derive synonyms which can be used in text mining and extraction approaches to link literature mentions of proteins and genes to network entities thereby enhancing functional information. n this study we investigated whether and with which accuracy functional descriptions and synonyms can be transferred between species in order to increase the number of annotated proteins. We propose a method to evaluate the quality of ortholog relations with protein functional descriptions and to transfer ortholog relations transitively along a given phylogenetic tree.

Johannes Raffler, Karsten Suhre
A genome-wide association study of metabolic traits in human urine Downoad paper
Abstract: We present a genome-wide association study of metabolic traits in human urine, designed to investigate the detoxification capacity of the human body. Using NMR spectroscopy, we tested for associations between 59 metabolites in urine from 862 male participants in the population-based SHIP study. We replicated the results using 1,039 additional samples of the same study, including a 5-year follow-up, and 992 samples from the independent KORA study. We report five loci with joint P values of association from 3.2 × 10−19 to 2.1 × 10−182 . Variants at three of these loci have previously been linked with important clinical outcomes: SLC7A9 is a risk locus for chronic kidney disease, NAT2 for coronary artery disease and genotype-dependent response to drug toxicity, and SLC6A20 for iminoglycinuria. Moreover, we identify rs37369 in AGXT2 as the genetic basis of hyper-β -aminoisobutyric aciduria.

Rayna Stamboliyska, Cedric Cagliero
Adaptation to osmotic stress in E. coli: old song, new melody Downoad paper
Abstract: Bacteria encounter widely varying environmental conditions and increase in salinity is one of the most frequent ones. Adaptation to this case, which is referred to as hyperosmotic stress, involves a modification of transcription patterns with downstream effects on physiology. Moreover, the nucleoid structure is highly sensitive to these changes (supercoiling has been reported) and to global gene expression through RNA polymerase binding and distribution. Here, we probed the impact of hyperosmotic stress (0.5 M NaCl) on the nucleoid structure of Escherichia coli K-12 coupled with a detailed survey of RNA polymerase binding using microscopy and ChIP-on-chip, respectively. Our observation showed surprising dynamics of the E. coli chromosome that appears consistent with the observed RNA polymerase distribution. Interestingly, RNA polymerase binding events appear to be less frequent during the stress period while the nucleoid shows global expansion. Furthermore, we assessed the transcriptional changes underlying the response to hyperosmotic stress and observed the activity of previously reported and unknown-to-date genes with respect to time. As previously reported, we did not observe any significant change in expression of rpoS, that is the gene encoding the osmotic stress sigma factor. Lastly, we reconstructed the regulatory interactions governing this adaptation and established the preferential involvement of simple transcriptional motifs. Our study thus demonstrates the crucial involvement of the RNA polymerase in the dynamics and topology of the bacterial chromosome and represents the first complete and comprehensive map of the events directing the adaptation of E. coli to increased salinity in the medium.

Pegah Tavakkolkhah, Ralf Zimmer, Robert Küffner
Time-sensitive inference of gene regulatory networks Downoad paper
Abstract: Many algorithms were devised to deduce gene regulatory networks (GRN) from mRNA expression data. Candidate transcription factor:target gene (TF:TG) relationships are assumed more likely if the expression of the TG depends on the expression of the TF. This dependency can for instance be evaluated by Pearsons linear correlation coefficient ρ2 or by η2, a non-parametric, non-linear correlation coefficient computed from an analysis of variance (ANOVA). In particular, η2 performed significantly better than previously published methods in the recent DREAM5 competition. Inference algorithms usually neglect to analyze whether expression changes in TFs precede expression changes in TGs. We present a simple but effective approach to extend standard algorithms (exemplified by ρ2 and η2) by an analysis of time shifted expression patterns from time series data and report the achieved performance improvements.