Multipoint linkage analysis using affected relative pairs and partially informative markers

Linkage analysis is a method of identifying regions of the human genome harboring genes affecting the risk for a particular disease. It works by finding chromosomal segments inherited by affected relatives from a common ancestor (i.e., identical by descent or IBD) in excess of that expected by chance. Two complicating factors are that only a relatively small number of genomic locations (marker loci) are examined and the number of distinct realizations (alleles) at each marker is not large. Hence, unambiguous determination of IBD is impossible for any genomic location without additional information. Assuming data from a set of mapped, partially informative markers, we evaluate the effectiveness of a method that analyzes the array of markers on each chromosome jointly (multipoint methods) as a function of the informativeness and density of the markers. For the special case of pairs of half siblings whose parents are also typed, a combination of analysis and simulation is used to obtain insight into the problem of setting thresholds to control the false-positive error rate. Approximations are given for the power, and guidelines are developed to help describe the trade-offs between marker density and informativeness.