When should I use an explicit background set in evaluating enrichments?

Statistical enrichments are always relative to some background set. If you do not provide a background set, then GREAT uses a whole genome background. Such a background is not always appropriate. You should provide a background set to account for any inherent biases in sampling 1.

References

Rhee S. Y. et al. Use and misuse of the gene ontology annotations, Nat Rev Genet. 9:509-515 (2008).