Table 1.

Resulting ten best haplotype patterns after replication of the multi-locus stepwise regression.

No.	SNP 1	SNP 2	SNP 3	SNP 4	MSR p value	MSR p value (corrected)	TDT p value	TDT p value (corrected)
1	rs7550106	rs499697	rs17659389	rs17670505	1.30 × 10 -13	4.13 × 10 -7	0.00013	0.03976
2	rs499697	rs17670505	rs576941	rs16835086	2.99 × 10-10	9.54 × 10-4	0.00033	0.09809
3	rs13373771	rs499697	rs6702463	rs17670505	9.85 × 10-10	3.14 × 10-3	0.00034	0.10260
4	rs6678672	rs4845766	rs499697	rs17659389	9.70 × 10-10	3.10 × 10-3	0.00040	0.11999
5	rs6678672	rs4845766	rs13373771	rs17659389	1.05 × 10-09	3.34 × 10-3	0.00051	0.15153
6	rs7550106	rs499697	rs989834	rs17670505	4.20 × 10-10	1.34 × 10-3	0.00053	0.15796
7	rs11204897	rs13373771	rs6701221	rs17670505	2.51 × 10-10	8.01 × 10-4	0.00055	0.16504
8	rs499697	rs17659389	rs17670505	rs16835086	4.77 × 10-10	1.52 × 10-3	0.00058	0.17369
9	rs2999547	rs499697	rs17659389	rs17670505	1.65 × 10-10	5.25 × 10-4	0.00061	0.18254
10	rs13373771	rs1923508	rs6701221	rs17670505	7.36 × 10-10	2.35 × 10-3	0.00074	0.22309

Multi-locus stepwise regression (MSR p value) was used for case control set (n = 1,914) and replicated in family set (268 families) using the weighted TDT statistic. Bonferroni corrections were done by multiplying 3188010 to the MSR p value and 300 to the TDT p value. Only the first haplotype pattern kept significant after Bonferroni correction.