How to incorporate interaction between two alleles in ApoE for Alzheimer's polygenic risk score?

Question

[kl]

Hi,

I am generating a polygenic risk score for Alzheimer's disease and want to include ApoE. However, I'm not sure how I'd include the interaction between these alleles in the PRS. For example, I include the two SNPs tagging the locus rs7412 and rs429358 with the effects provided by IGAP. However, in some cases this will overestimate the odds ratio conferred by them since its an additive model.

[Kevin Blighe]

What about creating a merged variable for rs7412 and rs429358 (4 possible values, considering 2 genotypes at each SNP)? That way, ApoE and the merged variable would only be in the model formula once.