Fig. 2From: A strategy for evaluating pathway analysis methodsThe procedure used to compute two metrics, (a) recall and (b) discrimination. Recall provides a measure of consistency of the overlap between the pathways identified from the original dataset with those obtained from sub-datasets randomly sampled from the original dataset. The computation of discrimination employs two datasets from different studies and multiple sub-datasets resampled from each of the two datasets. Discrimination measures the fraction of sub-datasets that yield higher recall with the original dataset from which they were resampled in comparison to the recall for another randomly selected datasetBack to article page