Genotype-driven identification of a molecular network predictive of advanced coronary calcium in ClinSeq® and Framingham Heart Study cohorts

Table 3 Predictive importance values of the set of SNPs that generate optimal predictive performance in both cohorts. Nearest genes are listed for intergenic SNPs (marked with asterisk)

SNP	Locus	Predictive importance	Predictive importance	Percent
		(ClinSeq®;)	(FHS)	difference
rs13159307	FBXL17 ^*	28.83	21.64	24.94
rs8107904	EMR2 ^*	36.95	21.83	40.92
rs571797	NRG3	17.68	6.86	61.20
rs2390285	MACC1	22.86	17.27	24.45
rs342393	NRG3	18.04	15.34	14.97
rs13429160	LOC101927701	35.68	16.89	52.66
rs11674863	LOC101927701	26.18	15.74	39.88
rs514237	NRG3	19.09	24.81	23.06
rs6860493	NNT	20.72	26.39	21.49
rs10054519	C5orf28	21.17	25.25	16.16
rs12521249	PAIP1 ^*	21.17	25.44	16.78
rs10065689	NNT	20.45	25.55	19.96
rs2241097	TLR5	34.02	24.11	29.13
rs10059993	NNT-AS1	20.82	24.77	15.95
rs12645809	ANTXR2	22.1	25.33	12.75
rs480220	NRG3	19.76	24.01	17.70
rs1366410	NNT	21.15	23.77	11.02
rs11767632	YAE1D1 ^*	32.09	20.94	34.75
rs7713479	NNT-AS1	21.11	37.48	43.68
rs243172	FOXN3	34.90	46.17	24.41
rs243170	FOXN3	35.91	51.20	29.86

The normalized difference of the predictive importance values of each SNP in two cohorts (difference divided by the higher predictive importance value in the two cohorts) has a median value of 24% (interquartile range:17%-36%). In terms of predictive importance based ranking, five of the top 11 SNP predictors (with 65% of the cumulative predictive importance) are common, whereas nine of the top 14 SNP predictors (with 76% of the cumulative predictive importance) overlap between two cohorts
^*Intergenic SNPs for which the nearest genes are reported

ISSN: 1752-0509