Efficient and biologically relevant consensus strategy for Parkinson’s disease gene prioritization

BMC Medical Genomics

Table 7 Statistical validation of the different gene prioritization strategies employed in this work (independently and in combination). Hypergeometric test, random bootstrap sampling experiment and enrichment features of the different gene prioritization strategies

	Hypergeometric Test			Random Bootstrap Sampling (100 Generations)						Enrichment
	n	m	p-value	Mean	Median	Min.	Max.	Std. Dev.	p-value (W)	Fold-Enrichment	TP Rate
Limma	134	10	0.0295	5.0410	5	0	17	2.1852	<0.0001	1.9837	0.0746
ML	168	11	0.0520	6.3211	6	0	22	2.4421	<0.0001	1.7402	0.0655
ML ∪ Limma	246	14	0.0805	9.2609	9	0	25	2.9426	<0.0001	1.5117	0.0569
PD_02	1437	73	0.0034	55.4259	55	25	87	6.6392	<0.0001	1.3171	0.0508
PD_07	494	26	0.0512	18.5957	18	2	41	4.1038	<0.0001	1.3982	0.0526
PD_02 ∪ PD_07	1931	99	0.0003	72.6709	73	37	112	7.3516	<0.0001	1.3623	0.0513
Concensus	50	7	0.0025	1.8817	2	0	10	1.3407	<0.0001	3.7200	0.1400

n: number or genes in the prioritized PD module; m: number of known PD related genes in GAD found in the prioritized module; p-value: hypergeometric probability of finding by chance k or more known PD related genes in a set of n prioritized genes; Mean/Median/Min./Max./Std. Dev.: average/median/minimum/maximum/standard deviation of the number of known PD related genes in GAD included in randomly selected gene sets with the same number of genes as the corresponding set of prioritized genes; Fold-enrichment: fold difference between m and Mean (Fold-enrichment = m/Mean); TP Rate: ratio of known PD related genes in n (TP Rate = m/n)

ISSN: 1755-8794