Overview of the analysis design. (a) First part of the analysis including quality controls, normalization, preliminary clustering performed on the untreated dataset, computation of the cluster centroids on the tamoxifen treated dataset, and estimation of signature stability with regards to signature size, using cross-validation. (b) Second part of the analysis including the classifier development, performance assessment by cross-validation and performance assessment on independent validation data sets.
Loi et al. BMC Genomics 2008 9:239 doi:10.1186/1471-2164-9-239