A novel approach to estimating parameters of the polygenic model has been proposed, showing promise for inferring the explained genetic variance and/or proportion of null markers. The method yields estimates that are similar to those obtained by existing approaches [1], [37]. A similar approach to estimation has been developed by Stahl et al [9], based on simulating GWAS data from proposed models, and using rejection sampling to construct posterior distributions of their parameters. Apart from the accommodation of prior distributions (which were uninformative), this is essentially the same approach as used here except that whole genome simulation is used to obtain a sampling distribution. The analytic results provided here should allow this approach to be implemented more efficiently, and this will be attempted in future work.
