Fig. 5

Evaluation of individual methods on the GEO2KEGG benchmark set. Depicted are competitive rank distributions of the KEGG target pathways according to (a) gene set p-value and (b) gene set score of the individual methods when applied to the 27 GEO datasets. The x-axis on top of both plots indicates the number of datasets for which the corresponding method resulted in the best ranking (among methods) of the target pathway. As an example, the leftmost blue boxplot in (a) shows for ORA a median rank of ≈20 %, i.e. ORA returned for half of the datasets a competitive rank below 20 %. Depicted on top, #best=7 means that ORA returned for 7 datasets the best ranking (among methods) for the target. Detailed rank distributions for each dataset can be found in Additional file 1: Figure S3 and S4