Table 2. Top case-only interaction results (P-value < 0.005 and OR ≥2).

Locus 1			Locus 2			Case-only test		Regression test
Gene	Marker	Chr:pos	Gene	Marker	Chr:pos	Odds ratio (95% CI)	P-valuea	Odds ratio	P-value	Full model pseudo-R2 b
ADRA1B	rs11953285	5:159 324 389	DBH	rs1611115	9:135 490 336	4.56 (2.10–20.02)	0.000054	4.83	0.0004	0.018
SLC6A3	Intron 8 VNTR	5:1 464 855	SNAP25	rs362562	20:10 191 186	2.41 (1.46–4.23)	0.000558	1.60	0.0871	0.008
ADRA1B	rs7737796	5:159 302 007	SNAP25	rs363006	20:10 228 083	2.43 (1.50–4.37)	0.000711	2.27	0.0102	0.016
ADRA1A	rs472865	8:26 754 388	DRD4	exon 3 VNTR	11:629 989	3.64 (1.61–14.41	0.00100	3.80	0.0048	0.015
ADRA1B	rs6884105	5:159 281 021	SNAP25	rs363006	20:10 228 083	2.18 (1.38–3.85)	0.00104	2.13	0.0215	0.008
ADRA1A	rs486354	8:26 766 207	DRD4	120 bp duplication	11:625 946	2.46 (1.42–5.21)	0.00134	2.66	0.0111	0.012
DBH	rs129882	9:135 513 490	SLC18A2	rs2283135c	10:118 989 922	7.27 (N/A)	0.00178	9.45	0.0037	0.016
DRD4	rs12720373	11:626 688	CES1	rs3815589	16:54 384 420	2.47 (1.38–5.36)	0.00204	2.21	0.0222	0.015
DRD5	D4S2928c	4:10 612 311	DRD4	exon 3 VNTR	11:629 989	5.84 (N/A)	0.00250	3.66	0.0664	0.007
CES1	rs3815589	16:54 384 420	ADRA1D	rs4815670	20:4 164 864	2.03 (1.23–3.47)	0.00314	2.03	0.0252	0.010
ADRA1A	rs472865	8:26 754 388	DBH	rs2797849	9:135 491 762	2.33 (1.36–4.71)	0.00328	2.28	0.0220	0.011
ADRA1B	rs13162302	5:159 299 853	ADRA1D	rs6052456	20:4 173 573	2.67 (1.41–6.64)	0.00410	3.82	0.0010	0.021
DRD5	D4S2928c	4:10 612 311	COMT	rs4680	22:18 331 271	2.60 (1.38–5.27)	0.00416	2.04	0.0904	0.004
HTR1B	rs1228814	6:78 230 539	SLC6A4	rs1872924	17:25 570 472	2.06 (1.23–4.09)	0.00456	3.24	0.0068	0.016
SLC18A2	rs363279	10:119 026 615	SLC6A2	rs9930182	16:54 298 798	3.38 (1.59–13.27)	0.00474	4.15	0.0033	0.017

VNTR, variable number of tandem repeats polymorphism; chr:pos, chromosome:position (base pairs from pter; hg18).

Odds ratio were rescaled to lie above 1, confidence intervals were derived from 1000 bootstrap samples, ‘N/A' indicates that a ceiling or floor effect was observed in the bootstrap estimates owing to sparse contingency table; see Supplementary Materials for the full results.

Note: the P-values presented are unadjusted for multiple testing; none of them are significant at P<0.05 after correction for the number of tests.

^aEmpirical P-value from 50 000 or 1 000 000 (in bold) Monte Carlo permutations.

^bMcFadden pseudo-R² for the full model including main effects and interaction.

^cMinor allele frequency < 0.1 in parents.