I am just designing a GWAS analysis that will be performed using an R package such as GenABEL. Our format will be a case-control study of women who did or did not develop gestational diabetes during pregnancy (not that that is important!).

As well as doing a standard GWAS additive model analysis, however, we are also interested in doing a separate analysis related to imprinted genes where we are interested in analysing associations with the alleles transmitted from the mother to her (unborn) child. We will only have SNP genotypes from mothers (i.e. not DNA family trios) so for each SNP in the GWAS we plan to analyse just those women where their genotype is homozygous (so that we can be sure which allele the mother transmitted to the baby). Could anybody recommend the easiest way of doing this? Would it be easiest to write some R code to strip the heterozygote genotypes from the database, and then do a pseudo-additive model analysis?