Table 2. Variable selection results.

Genomic regiona	Locationb	Phenotype	SNPs	Estimate	Individual P-valuec	Overall P-valued
F12 and GRK6	5q33-35	aPTT	rs2545801 rs1801020	−0.02950 −0.022793	0.041 0.1303	< 2.2 × 10−16
HRG and KNG1	3q27	aPTT	rs1042445	0.010413	0.00704	< 2.2 × 10−16
			rs2062632	0.007951	0.02904
			rs2228243	0.018532	2.34 × 10−05
			rs710446	−0.021965	2.96 × 10−12
			rs9898	−0.014528	6.23 × 10−05
KLKB1	4q35	aPTT	rs4253304	0.015523	6.15 × 10−08	6.15 × 10−08
ABO	9q34	aPTT	rs657152	0.015269	0.000263	1.269 × 10−10
			rs651007	0.006843	0.154113
F5 and SLC19A2	1q23	APC resistance	rs6025	0.28882	<2 × 10−16	< 2.2 × 10−16
HRG	3q27	APC resistance	rs16860992	−0.116523	0.043113	2.2 × 10−16
			rs2228243	−0.088207	0.125584
			rs9898	−0.014962	0.000618

^aGenomic regions are represented by gene names.

^bLocation is cytogenetic location.

^cThe individual P-value gives an indication of the statistical independence of a SNP from others included in the model.

^dThe overall P-value indicates the strength of the combined evidence against the null hypothesis when all SNPs are included.

For each SNP in a genomic region the estimate gives an indication of the independent effect of that SNP on the trait. Akaike Information Criterion was used to identify SNPs with a significant independent contribution to the phenotype of interest in a step-wise multiple regression of significantly associated SNPs in each genomic region.