Table 3.

Conditional analysis of (a) BANK1 and (b) TNFSF4

	Logistic regression			Haplotype-based association test
	Controlled SNP	Effective SNP	P-values	Effective SNP	P-values
(a)
GWAS	rs4522865	rs10516487	0.042	Omnibus	9.06 × 10−3
	*rs10516487	rs4522865	1.03 × 10−3	rs4522865	0.031
				rs10516487	0.047
Replication	*rs17266594	rs3733197	0.037	Omnibus	1.29 × 10−8
	rs3733197	rs17266594	6.63 × 10−8	rs3733197	0.028
				rs17266594	2.01 × 10−8
(b)
GWAS	rs2205960	rs844648	0.467	Omnibus	3.12 × 10−5
	rs2205960	rs844644	0.361	rs2205960	3.21 × 10−3
				rs844648	0.523
Replication	rs2205960	rs844648	0.551	Omnibus	3.81 × 10−4
	rs844648	rs2205960	6.26 × 10−3	rs2205960	5.16 × 10−3
				rs844648	0.489

Abbreviations: GWAS, genome-wide association study; SNP, single nucleotide polymorphism.

Conditional analysis of SNPs from BANK1 and TNFSF4 in both GWAS and TaqMan replication data using PLINK.

‘^*', rs10516487 and rs17266594 have strong LD in Chinese (r²=1 in HapMap-HCB).

SNPs chosen for analysis were the five top-rank SNPs from our GWAS data in the BANK1 and TNFSF4 regions, respectively, and SNPs with reported association with SLE in Caucasian studies, as bolded in Table 1 and labeled in Figure 1. In the section of ‘logistic regression', the column marked by ‘controlled SNP' lists the SNPs whose effects were controlled whereas the column marked by ‘effective SNPs' lists the SNPs in concern during pairwise logistic regression. P-value is for the individual effect of the effective SNP. Only SNPs with significant P-values during logistic regression were shown, except rs844648 and rs844644 in TNFSF4. In the section of ‘haplotype-based association test', haplotypes were constructed for BANK1 and TNFSF4 correspondingly with the selected SNPs bolded in Table 1. The column marker by ‘effective SNP' contains SNPs whose independent effects were in concern and ‘P' represents the P-values for the significance of ‘effective SNP' when the effects of the other SNPs within the same haplotype were controlled.