Additional file 5: of Identification and validation of single-sample breast cancer radiosensitivity gene expression predictors

Figure S1. Selection of top discrimination genes in the Illumina discovery cohort data. Number of genes in the random forest models are plotted against performance of classifying cases and controls, as measured by cross-validated area under the curve (AUC). The analysis was stratified for estrogen receptor (ER) status and radiotherapy (RT) treatment, and with added patients from other strata, based on a biological rationale as described in the text. (ZIP 171 kb)