
Conditional haplotypebased association test, P value


Sx ^{b}

Independent effect of Sx ^{c}

Controlling for Sx ^{d}


rs12342421 (S8)

0.0005

0.4360

rs10974944 (S9)

– ^{e}

0.0072

rs10119004 (S10)

0.4700

2.84 × 10^{−7}

rs10974947 (S11)

0.2480

1.79 × 10^{−14}

rs12343867 (S12)

0.7970

0.0019

rs12340895 (S13)

– ^{e}

0.0072

 ^{a}This table shows the individual effects of the constituent single nucleotide polymorphisms (SNPs) on the shortest and most significant sliding window that contains the most impressive SNP in our study (rs12342421, i.e. S8) and the riskhaplotypetagging SNPs (rs10974944, rs12343867 and rs12340895, i.e. S9, S12 and S13). Conditional logistic regression was performed with adjustment for sex and age. The shortest and most significant sliding window carrying these four SNPs is S8…S13 (see Table 2). The conditional omnibus test invoked by the “chap” command of PLINK gives a P value of 1.34 × 10^{−14} (based on likelihood ratio test).
 Note that this P value is similar, but not identical, to the P value of 2.75 × 10^{−12} (based on Wald test, Table 2) generated by the omnibus test of logistic regression invoked by the “logistic” command of PLINK in the slidingwindow approach.
 ^{b}Sx indicates the SNP tested for an independent effect one at a time by the conditional haplotypebased analysis of the sliding window S8…S13. Please refer to Table 1 for the identity of the SNPs concerned.
 ^{c}Omnibus P value for the effect of Sx that is independent of the other SNPs in the sliding window S8…S13.
 ^{d}Omnibus P value for the sliding window S8…S13 when Sx is controlled for.
 ^{e}Not a valid comparison due to identical alternate and null models