Analysis of genome-wide association study data using the protein knowledge base

BMC Genomic Data

Table 2 Average number of gene predictions made by Gentrepid modes and modules

Mode	SNP/gene		CPS				CMP				CPS + CMP
	Approach		WS	MWS	MHS	HS	WS	MWS	MHS	HS	WS	MWS	MHS	HS
Seeded	BY	1 Mbp	29.14	6.00	2.29	0.71	15.43	3.86	0.71	0.43	42.86	9.43	3.00	1.14
		0.5 Mbp	12.29	5.29	2.00	1.29	9.00	2.86	0.57	0.29	20.71	7.71	2.57	1.57
		0.1 Mbp	6.71	3.00	1.71	1.00	3.14	1.00	0.29	0.00	9.57	3.57	1.86	1.00
	NN	Adjacent	19.14	6.00	1.57	1.43	6.86	1.57	0.43	0.00	25.29	7.00	1.86	1.43
		Nearest	10.86	2.57	1.43	1.00	3.00	0.86	0.29	0.00	13.71	3.29	1.57	1.00
		Resident	4.29	1.43	1.00	0.57	1.29	0.43	0.14	0.00	5.57	1.86	1.14	0.57
Ab initio	BY	1 Mbp	105.00	27.14	6.86	5.86	57.57	14.71	12.57	10.14	157.14	40.71	17.43	14.29
		0.5 Mbp	41.29	17.43	5.86	2.86	30.71	12.71	8.57	10.43	70.00	29.57	14.43	11.86
		0.1 Mbp	28.57	6.71	2.00	0.57	12.14	10.00	7.86	3.57	38.14	16.71	9.86	4.14
	NN	Adjacent	59.57	13.86	2.14	0.71	26.86	8.71	5.29	2.00	81.14	21.29	6.71	2.71
		Nearest	28.71	5.14	1.00	0.57	10.86	2.00	1.71	0.00	37.43	6.86	2.71	0.57
		Resident	13.00	2.57	0.29	0.00	9.71	0.86	0.57	0.00	21.00	3.14	0.86	0.00
Seeded and ab initio	BY	1 Mbp	105.00	27.14	6.86	6.57	72.43	18.43	13.29	10.57	170.14	44.00	18.14	15.43
		0.5 Mbp	41.43	17.43	5.86	3.57	39.57	15.57	9.14	10.71	78.43	32.00	15.00	12.86
		0.1 Mbp	28.57	7.00	2.29	1.29	15.29	11.00	8.14	3.57	40.57	17.57	10.29	4.86
	NN	Adjacent	59.57	13.86	2.29	1.71	33.43	10.29	5.71	2.00	86.57	22.29	7.14	3.71
		Nearest	28.71	5.29	1.71	1.14	13.86	2.86	2.00	0.00	39.86	7.57	3.57	1.14
		Resident	13.00	2.71	1.00	0.57	10.71	1.29	0.71	0.00	22.00	3.71	1.71	0.57

Each cell represents the number of predictions averaged across the seven phenotypes made by the mode and module. Column abbreviations: HS, highly significant; MHS, moderately-high significance; MWS, moderately-weak significance; WS, weakly significant; CPS, common pathway scanning module; CMP, common module profiling. Row abbreviations: Seeded, mode using known disease gene information as seeds; Ab initio, blind approach mode; BY, bystander approach; NN, nearest neighbour approach. The total number of predictions made by Gentrepid for each significance threshold and SNP/gene approach are bolded. All results have been filtered on significance. For CPS, pathways reach a significance threshold of p < 0.05 based on Fishers exact test. For CMP seeded, a threshold of 0.4. For CMP ab initio, a threshold of χ²_{max_unique} > 105 and χ^{2 min} > 7.88 for multidomain proteins and χ^{2 min} > 100 for single domain proteins.

ISSN: 2730-6844