Effect of data set and classifier gene number on survival.

The effect of training data set and classifier gene selection was assessed on a previously unseen data set GSE4475 (A). Survival separation observed with the published GEO classes (top) was compared to the meta-classifier (middle) and the LPS classifier (bottom). The Hazard Ratio (GCB vs ABC as baseline), 95% confidence interval of the Hazard Ratio, and the resulting p-value for survival separation are shown. For the meta-classifier and LPS the results for training on the Wright et al. data set (20 classifier genes) are shown, followed by the results for classifiers trained on the GSE10846 R-CHOP data set with different sets of classifier genes (Wright20-Wright5, Dave185-Dave10, All185-All10). (B) The classifiers were also tested on Monti et al. data requiring MaxAvgMerge data processing (see Materials and Methods) to observe significant differences in survival. Results are shown for the LMT_J48_RF100_SMO and LPS (best shown) classifiers trained on Wright et al. data and using 20 classifier genes.