For each disease-SNP association, five tests were performed (Table 1). A standard case-control test without taking parental origin into account was performed to provide a baseline. Then a case-control analysis was performed separately for the paternally and maternally inherited alleles. A two-degree of freedom test was applied to evaluate the joint effect. A multiplicative model was assumed for the two alleles, but the magnitude and direction of the effect were allowed to differ. Finally, the difference between the effects of the paternally and maternally inherited alleles was directly tested by comparing their allele frequencies within cases. The information for this test mainly comes from the counts of the two types of heterozygotes within cases. (see Supplementary Information for details).
