We show that we can use conditional p-values to deal with multiple independent associations with the same trait at one locus. While we found this solution generally effective, Wallace [14] points out that this top SNP selection for the conditional analysis can create biases, although the bias is small in the case of large samples and/or strong effects. For difficult loci with multiple associations for both traits and available genotype data, it may be more appropriate to estimate Bayes factors for sets rather than single variants in order to obtain an exact answer. This extension would avoid the issue of SNP selection for the conditional analysis.
