The use of multiple restriction fragment length polymorphisms in prenatal risk estimation. I. X-linked diseases.

An analytical procedure for estimating the risk of X-linked diseases based on presence/absence of a series of restriction sites is presented. Multiple-locus linkage phase of the carrier mother is first inferred from previous offspring, from parents, and by molecular means. Bayesian risk estimates are then obtained using this information and the recombination-segregation distribution. The improvement afforded by using multiple flanking markers rather than a single marker is dramatic. Whereas the upper bound on the probability that a family will be informative using a single diallelic X-linked marker is .5, in the case of m markers, the bound on the probability of an informative family becomes 1 - .5m. With a single linked marker, the precision in the risk estimate is bounded by the frequency of recombination, whereas the requirement of very tight linkage is relaxed somewhat when multiple flanking markers are used. Recombination interference and multiple-locus linkage disequilibria can further improve the risk estimates, but it is important to understand how the statistical confidence in these parameters affects the reliability of the risk estimates.