Table 6.

Number of codependencies detected at a significance cutoff of P < 0.05 without multiple hypothesis correction or after one of three methods for multiple hypothesis testing correction (Bonferroni, Benjamini–Hochberg, and Benjamini–Yekutieli).

	No prior	INDRA prior
		Total	INDRA only
Number of comparisons (non‐mitochondrial)	121,778,711	265,874	N/A a
Correlations with uncorrected P < 0.05	21,526,511	63,926 a	N/A a
Significant corrs after Bonferroni	99,544	4,982	1,836
Significant corrs after Benjamini–Hochberg	5,025,535	30,127	7,506
Significant corrs after Benjamini–Yekutieli	972,831	12,812	4,007

Results are shown for a case in which no prior is used and data are analyzed directly (“No prior”), or when an INDRA prior is used (“INDRA prior/Total”). The rightmost column shows the number of novel codependencies recovered exclusively when an INDRA prior was used along with correction for multiple testing (“INDRA prior/INDRA only”).

^a Figures for uncorrected P‐values do not apply to the “INDRA prior/INDRA‐only” case because without correction for multiple testing, the prior does not play a role in determining significance. Figures are shown for the “INDRA prior/Total” case to establish the number of codependencies with uncorrected P‐values > 0.05 that fall within the scope of the INDRA network; this serves as an upper bound for the number of correlations determined to be significant with the different approaches to multiple testing shown in the bottom three rows.