Table 5: Assessing FDR control.

Each p-value comes from a single t-test measuring whether the mean of the estimated FDP over the 10 sub-runs is significantly larger than the selected 5% FDR threshold. Each column uses a different computationally concatenated UPS1 and yeast run, and each row refers to a different FDR estimation procedure. Boldface values are significant at a Bonferroni corrected threshold of 0.004 (0.05/12). The analysis suggests that all-sub fails to control FDR.

FDR control method	p-value 1	p-value 2	p-value 3
subset-search	0.25695	0.02642	0.20952
all-sub	0.0012944	0.0008953	0.0024908
group-FDR	0.05458	0.04435	0.05003
subset-neighbor search	0.59906	0.04705	0.56590