Principal-component-based population structure adjustment in the North American Rheumatoid Arthritis Consortium data: impact of single-nucleotide polymorphism set and analysis method

Table 3 Results for positive control SNPs for six methods of adjusting for structure

	rs2476601 [11, 12]		rs3761847 [1]
Adjustment Method	β^a ± SE	p-Value	β^a ± SE	p-Value
A. Logistic regression adjusting for DRB	0.61 ± 0.11	3.70 × 10^-8	0.42 ± 0.07	8.04 × 10^-9
B. Logistic regression adjusting for DRB and PC1-4	0.47 ± 0.12	1.32 × 10^-4	0.39 ± 0.08	2.08 × 10^-6
C. Logistic regression adjusting for DRB and cluster	0.46 ± 0.12	1.57 × 10^-5	0.45 ± 0.08	4.21 × 10^-8
D. Logistic regression adjusting for DRB, PC1-4, and cluster	0.46 ± 0.13	2.44 × 10^-4	0.42 ± 0.08	4.55 × 10^-7
E. PSAT based on PC1-4 cluster	-----^b	1.17 × 10^-6	-----	2.80 × 10^-8
F. Genomic control	0.16 ± 0.12	7.44 × 10^-7	0.42 ± 0.08	2.22 × 10^-7
Previously reported	0.50 ± 0.15	6.60 × 10^-4	0.35 ± 0.04	4.00 × 10^-14

^aβ, log-odds ratio for disease for the minor allele estimated by the logistic regression
^b---, Estimates of effect cannot be obtained.

ISSN: 1753-6561