Selecting maximally informative sibships for QTL linkage analysis

Abstract

Much work has been done in the area of increasing power of linkage studies by use of selective sampling. It is generally known that maximally discordant sib pairs are most informative, although affected pairs can be more informative still in the presence of a rare recessive. However, little work has been done in the area of construction of an optimal sample, once the sample size for genotyping has been determined. For example, clearly even under a simple additive model, the most extremely affected pairs will be more informative than all but the most extremely discordant pairs. When conducting a large-scale study where all sibships have been phenotyped and the next task is to select the potentially most informative sibships or members of a sibship for genotyping, a method for rank ordering all the sibships in a study by their potential informativeness for linkage would be most desirable. This will describe such a method and its implementation in a freely distributed Fortran program.