BACKGROUND: Recent advances in antibody microarray technology have made it possible to measure the expression of hundreds of proteins simultaneously in a competitive dual-colour approach similar to dual-colour gene expression microarrays. Thus, the established normalisation methods for gene expression microarrays, e.g. loess regression, can in principle be applied to protein microarrays. However, the typical assumptions of such normalisation methods might be violated due to a bias in the selection of the proteins to be measured. Due to high costs and limited availability of high quality antibodies, the current arrays usually focus on a high proportion of regulated targets. Housekeeping features could be used to circumvent this problem, but they are typically underrepresented on protein arrays. Therefore, it might be beneficial to select invariant features among the features already represented on available arrays for normalisation by a dedicated selection algorithm. RESULTS: We compare the performance of several normalisation methods that have been established for dual-colour gene expression microarrays. The focus is on an invariant selection algorithm, for which effective improvements are proposed. In a simulation study the performances of the different normalisation methods are compared with respect to their impact on the ability to correctly detect differentially expressed features. Furthermore, we apply the different normalisation methods to a pancreatic cancer data set to assess the impact on the classification power. CONCLUSIONS: The simulation study and the data application demonstrate the superior performance of the improved invariant selection algorithms in comparison to other normalisation methods, especially in situations where the assumptions of the usual global loess normalisation are violated.

MOTIVATION: The functional interpretation of microarray datasets still represents a time-consuming and challenging task. Up to now functional categories that are relevant for one or more experimental context(s) have been commonly extracted from a set of regulated genes and presented in long lists. RESULTS: To facilitate interpretation, we integrated Gene Ontology (GO) annotations into Correspondence Analysis to display genes, experimental conditions and gene-annotations in a single plot. The position of the annotations in these plots can be directly used for the functional interpretation of clusters of genes or experimental conditions without the need for comparing long lists of annotations. Correspondence Analysis is not limited in the number of experimental conditions that can be compared simultaneously, allowing an easy identification of characterizing annotations even in complex experimental settings. Due to the rapidly increasing amount of annotation data available, we apply an annotation filter. Hereby the number of displayed annotations can be significantly reduced to a set of descriptive ones, further enhancing the interpretability of the plot. We validated the method on transcription data from Saccharomyces cerevisiae and human pancreatic adenocarcinomas. AVAILABILITY: The M-CHiPS software is accessible for collaborators at http://www.mchips.org

MOTIVATION: Microarray technology provides access to expression levels of thousands of genes at once, producing large amounts of data. These datasets are valuable only if they are annotated by sufficiently detailed experiment descriptions. However, in many databases a substantial number of these annotations is in free-text format and not readily accessible to computer-aided analysis. RESULTS: The Multi-Conditional Hybridization Intensity Processing System (M-CHIPS), a data warehousing concept, focuses on providing both structure and algorithms suitable for statistical analysis of a microarray database's entire contents including the experiment annotations. It addresses the rapid growth of the amount of hybridization data, more detailed experimental descriptions, and new kinds of experiments in the future. We have developed a storage concept, a particular instance of which is an organism-specific database. Although these databases may contain different ontologies of experiment annotations, they share the same structure and therefore can be accessed by the very same statistical algorithms. Experiment ontologies have not yet reached their final shape, and standards are reduced to minimal conventions that do not yet warrant extensive description. An ontology-independent structure enables updates of annotation hierarchies during normal database operation without altering the structure. AVAILABILITY AND SUPPLEMENTARY INFORMATION: http://www.dkfz.de/tbi/services/mchips

In physical mapping, one orders a set of genetic landmarks or a library of cloned fragments of DNA according to their position in the genome. Our approach to physical mapping divides the problem into smaller and easier subproblems by partitioning the probe set into independent parts (probe contigs). For this purpose we introduce a new distance function between probes, the averaged rank distance (ARD) derived from bootstrap resampling of the raw data. The ARD measures the pairwise distances of probes within a contig and smoothes the distances of probes across different contigs. It shows distinct jumps at contig borders. This makes it appropriate for contig selection by clustering. We have designed a physical mapping algorithm that makes use of these observations and seems to be particularly well suited to the delineation of reliable contigs. We evaluated our method on data sets from two physical mapping projects. On data from the recently sequenced bacterium Xylella fastidiosa, the probe contig set produced by the new method was evaluated using the probe order derived from the sequence information. Our approach yielded a basically correct contig set. On this data we also compared our method to an approach which uses the number of supporting clones to determine contigs. Our map is much more accurate. In comparison to a physical map of Pasteurella haemolytica that was computed using simulated annealing, the newly computed map is considerably cleaner. The results of our method have already proven helpful for the design of experiments aimed at further improving the quality of a map.

MOTIVATION: The technology of hybridization to DNA arrays is used to obtain the expression levels of many different genes simultaneously. It enables searching for genes that are expressed specifically under certain conditions. However, the technology produces large amounts of data demanding computational methods for their analysis. It is necessary to find ways to compare data from different experiments and to consider the quality and reproducibility of the data. RESULTS: Data analyzed in this paper have been generated by hybridization of radioactively labeled targets to DNA arrays spotted on nylon membranes. We introduce methods to compare the intensity values of several hybridization experiments. This is essential to find differentially expressed genes or to do pattern analysis. We also discuss possibilities for quality control of the acquired data. AVAILABILITY: http://www.dkfz.de/tbi CONTACT: [email protected]