Figure 6
Gene signatures are robust predictive biomarkers of exposure to radiation.
(a) Five-fold cross-validation of our 10 best predictive genes on the dataset of Paul and Amundson14 resulted in 97% accuracy for sample classification. (b) Five-fold cross-validation of the 10 best predictive genes from the dataset of Paul and Amundson14 as identified by the Random Forests model resulted in 100% accuracy for classification of our samples. (c) Overlap between differentially expressed genes from this study and gene signatures identified by Paul and Amundson6, Chauhan and co-authors18 and Warters and co-authors17. Common genes between all datasets are DDB2, POLH, MDM2, RPS27L, FDXR, CCNG1, TRIAP1, SESN1, FBXO22, PPM1D, ANKRA2, CDKN1A, TRIM22 and BBC3.
