Module-Based Outcome Prediction Using Breast Cancer Compendia

• DOI: 10.1371/journal.pone.0001047
• Publication Date: 2007-10-24T16:09:07Z
• Authors: Martin H. van Vliet; Christiaan N. Klijn; Lodewyk F. A. Wes...; Marcel J. T. Rein...
• Keywords: Compendium

Background The availability of large collections microarray datasets (compendia), or knowledge about grouping genes into pathways (gene sets), is typically not exploited when training predictors disease outcome. These can be useful since a compendium increases the number samples, while gene sets reduce size feature space. This should favorable from machine learning perspective and result in more robust predictors. Methodology We extracted modules regulated sets, compendia. Through supervised analysis, we constructed which employ predictive breast cancer To validate these applied them to independent data, same institution (intra-dataset), other institutions (inter-dataset). Conclusions show that derived single achieve better performance on validation data compared gene-based also there trend specificity performance: dataset, specific perform those human compendium. Additionally, module-based predictor provides much richer insight underlying biology. Frequently selected are associated with processes such as cell cycle, E2F regulation, DNA damage response, proteasome glycolysis. analyzed two related OCT1 transcription factor, respectively. On an individual basis, provide significant separation survival subgroups data.