Table 2. Significant GO term result overlaps of the 17 algorithms on the four case studies.

	Enrichments only		Corrected
	Average	SD	Average	SD	Down-sampling (5)	SD
Two or more methods
All results	58.4%	9.2%	52.4%	5.3%	24.2%	10.7%
Both robust & unique	37.4%	17.9%	29.0%	8.5%	15.9%	8.1%
Enriched across all studies (union)
All results	37.2%	4.1%	97.3%	1.1%	58.2%	22.9%
Both robust & unique	29.2%	4.8%	91.6%	4.9%	48.0%	7.9%
Enriched per study (averaged)
Both robust & unique	16.7%	10%	30.7%	26%	30.5%	6%