. 2021 Apr 19;8:635074. doi: 10.3389/fmolb.2021.635074

TABLE 5.

Co-mention validation of phenotype-function pairs. “All” corresponds to all pairs including a function from Table 4. “Confirmed” refers to the number of these pairs that were significantly co-mentioned in PubMed. “Random” refers to the number of pairs in a randomised list based on “All” pairs that were significantly co-mentioned in PubMed; average of 100 random datasets ± SD is shown in this case.

Paired	Phenotypes
function	OMIM			Orphanet
	All	Confirmed	Random ^a	All	Confirmed	Random ^a
GO	567 721	26 841	$10 814 \pm 80.5$	535 389	21 863	$9 613 \pm 93.4$
KEGG	17 679	1 858	$627 \pm 22.1$	17 104	1 556	$628 \pm 21.2$
Reactome	82 826	3 278	$1001 \pm 166.3$	78 409	2 402	$816 \pm 28.7$

^{^a}

The randomised phenotype-function pair set was formed by shuffling the links between the pairs in each list, keeping the total number of links per phenotype/function unchanged. This sampling procedure was repeated to produce 100 different replicas of randomised phenotype-function pairs. These sets were used in the corresponding Fisher’s exact tests.