A two-dimensional genome scan for rheumatoid arthritis susceptibility loci

BMC Proceedings

Table 1 Highest two-locus peaks on the 2D surface

			Two-locus results^a
Locus 1	Θ	Locus 2	GEN (microsatellite)	GEN-ADD P-value	GEN-MUL P-value	ε^b	GEN^c (microsatellite and SNP)	1-Lod unit support intervals (cM)^d
Entire 2D surface
6p21	0.5	16p12	18.02	0.011	0.011	10³	21.86	(51–55); (37–59)
6p21	0.5	18q21	17.90	0.254	0.937	0.66	23.04	(51–55); (52–85)
6p21	0.5	8p23	17.85	0.036	0.413	8.12	22.27	(51–55); (0–26)
1q42	0.5	6p21	17.44	0.642	0.083	0	21.58	(221–270); (51–55)
6p22	0.11	6p21	17.44	0.999	0.009	0	21.5	(31–47); (51–58)
2D surface excluding 6p21 interactions
8p23	0.5	18q21	4.60	0.631	0.587	0	3.24	(0–26); (52–85)
1p21	0.5	18q21	4.33	0.114	0.410	53	3.42	(126–153); (26–85)

^aTwo-locus general model MLS obtained in 511 families with microsatellite genotypes alone. Empirical P-values assessing the significance of the difference in two-locus MLS under the general and additive models, and the general and multiplicative models. Significant deviations from the null two-locus model (ADD or MUL) are shown in bold. Empirical P-values were also obtained assuming a null effect at the two loci using published thresholds [3], and for pairs of loci excluding 6p21 these were estimated at P = 0.34 for 8p23-18q21, and P = 0.53 for 1p21-18q21.
^bMaximum-likelihood estimate of epsilon.
^cTwo-locus general model MLS in 744 families using combined microsatellite and SNP data.
^d1-Lod unit support intervals (to nearest Rutgers Kosambi cM) obtained using the GEN statistic from the microsatellite data set.

ISSN: 1753-6561