EMP: Empirical correction of the significance of GO terms reported by active module identification (AMI) algorithms.

AMI algorihms receive a gene network and gene activity scores as input and report sub-networks (modules) that are active. A standard analysis of the modules is GO functional enrichment, which is used to interpret the results.

We observed that many AMI methods are prone to over-reporting of enrichment: GO terms enriched in the modules on real data were often also enriched when the algorithms were run on randomly permuted activity scores.

To correct this bias, we designed EMP (EMpirical Pipeline), a method that evaluates the empirical significance of GO terms reported as enriched in modules, by performing multiple runs on randomly permuted inputs.