Table 9. Genotype frequencies and effect sizes for a two-way interaction from a permuted data set that was significant at the 5% threshold with logistic regression.

	B
	0	1	2	Total
A
0	373/355	660/762	359/344	1392/1461
	1 (N/A)	1.21 (1.01–1.45)	1.01 (0.82–1.24)	1 (N/A)
1	183/102	294/291	148/135	625/528
	0.59 (0.44–0.78)	1.04 (0.84–1.29)	0.96 (0.73–1.26)	0.80 (0.70–0.92)
2	14/12	27/37	15/17	56/66
	0.90 (0.41–1.97)	1.44 (0.86–2.41)	1.19 (0.59–2.42)	1.12 (0.78–1.62)
Total
	570/469	981/1090	522/496
	1 (N/A)	1.35 (1.16–1.57)	1.15 (0.97–1.37)

Genotype frequencies and effect sizes for a two-way interaction from a permuted data set that was significant at the 5% threshold with logistic regression. Loci: A and B. At each locus: 0=common homozygote, 1=heterozygote and 2=rare homozygote. In each cell: Controls/cases; odds ratio (95 % confidence interval). The naïve P-value is 0.000044 (8 df) and the log-likelihood ratio statistic is 33.81. The data set was picked out based on the naïve P-value, which is the threshold naïve P-value for the 5% tail of the empirical null distribution.