Modelling linkage and association, with evaluation of common sampling schemes

Abstract

The probability model of parental transmission (of a multiallelic marker) to affected offspring has been reparameterized in such a way as to distinguish the recombination fraction parameter from the association parameters. The advantage of this reparameterization is that statistical tests can be developed which disentangle linkage and association. It is shown that these tests perform well compared with a standard approach that is commonly used. Further, the simulated data (Problem 2) enable a comparison to be made of the performance of these tests under different family sampling schemes.