. Author manuscript; available in PMC: 2015 Nov 1.

Published in final edited form as: Ann Hum Genet. 2014 Sep 17;78(6):478–491. doi: 10.1111/ahg.12080

Table 1.

False positive rate and proportion of variance explained by PCs using different summarizing strategies by different gene size under Simulation II with random support and quantitative traits. The genes are classified into small size (1~33 SNPs), medium size (34~67 SNPs) and large size (68~102 SNPs), hence there are six categories for the gene pairs: SS, SM, SL, MM, ML and LL. “PC1_SNP_×_SNP” indicates the first PC from the interaction design matrix and “PC1_geneS×PC1_geneT” indicates the cross product of the PCs for gene S and for gene T.

Size of Gene Pairs			SS	SM	SL	MM	ML	LL
Average Number of SNP× SNP Terms			84	443	923	2289	4800	9996
Proportion of variance explained		PC1_SNP_×_SNP	0.40	0.26	0.19	0.15	0.10	0.05
Proportion of variance explained		PC1_geneS×PC1_geneT	0.32	0.16	0.05	0.07	0.02	0.008
False Positive Rate	No size weight^*	PC1_SNP_×_SNP	0.002	0.004	0.005	0.01	0.038	0.083
	No size weight^*	PC1_geneS×PC1_geneT	0.003	0.006	0.034	0.023	0.076	0.100
	With size weight^**	PC1_SNP_×_SNP	0.003	0.005	0.005	0.009	0.025	0.058
	With size weight^**	PC1_geneS×PC1_geneT	0.004	0.008	0.036	0.021	0.053	0.073

“No size weight” indicates setting ω_size = 1; ω_effect and ω_path are set as proposed.

^**

“With size weight” indicates setting ω_size as proposed; ; ω_effect and ω_path are set as proposed.