BACKGROUND: Finding peaks in ChIP-seq is an important process in biological inference. In some cases, such as positioning nucleosomes with specific histone modifications or finding transcription factor binding specificities, the precision of the detected peak plays a significant role. There are several applications for finding peaks (called peak finders) based on different algorithms (e.g. MACS, Erange and HPeak). Benchmark studies have shown that the existing peak finders identify different peaks for the same dataset and it is not known which one is the most accurate. We present the first meta-server called Peak Finder MetaServer (PFMS) that collects results from several peak finders and produces consensus peaks. Our application accepts three standard ChIP-seq data formats: BED, BAM, and SAM. RESULTS: Sensitivity and specificity of seven widely used peak finders were examined. For the experiments we used three previously studied Transcription Factors (TF) ChIP-seq datasets and identified three of the selected peak finders that returned results with high specificity and very good sensitivity compared to the remaining four. We also ran PFMS using the three selected peak finders on the same TF datasets and achieved higher specificity and sensitivity than the peak finders individually. CONCLUSIONS: We show that combining outputs from up to seven peak finders yields better results than individual peak finders. In addition, three of the seven peak finders outperform the remaining four, and running PFMS with these three returns even more accurate results. Another added value of PFMS is a separate report of the peaks returned by each of the included peak finders.

MOTIVATION: Pre-selection of informative features for supervised classification is a crucial, albeit delicate, task. It is desirable that feature selection provides the features that contribute most to the classification task per se and which should therefore be used by any classifier later used to produce classification rules. In this article, a conceptually simple but computer-intensive approach to this task is proposed. The reliability of the approach rests on multiple construction of a tree classifier for many training sets randomly chosen from the original sample set, where samples in each training set consist of only a fraction of all of the observed features. RESULTS: The resulting ranking of features may then be used to advantage for classification via a classifier of any type. The approach was validated using Golub et al. leukemia data and the Alizadeh et al. lymphoma data. Not surprisingly, we obtained a significantly different list of genes. Biological interpretation of the genes selected by our method showed that several of them are involved in precursors to different types of leukemia and lymphoma rather than being genes that are common to several forms of cancers, which is the case for the other methods. AVAILABILITY: Prototype available upon request.

UNLABELLED: The Linnaeus Centre for Bioinformatics Data Warehouse (LCB-DWH) is a web-based infrastructure for reliable and secure microarray gene expression data management and analysis that provides an online service for the scientific community. The LCB-DWH is an effort towards a complete system for storage (using the BASE system), analysis and publication of microarray data. Important features of the system include: access to established methods within R/Bioconductor for data analysis, built-in connection to the Gene Ontology database and a scripting facility for automatic recording and re-play of all the steps of the analysis. The service is up and running on a high performance server. At present there are more than 150 registered users. AVAILABILITY: An open functional version is available at https://dw.lcb.uu.se/index.phtml?i_login=test. User accounts are created upon request. Additional facilities including plug-ins, user documentation and a password protected data storage system are available from http://www.lcb.uu.se/lcbdw.php