Abstract
Mean and maximum lod scores were generated by computer simulation methods (SIMLINK) to examine relative informativeness for linkage of 12 pedigrees with three-five affected members, assuming linkage of a rare (population allele frequency = 0.01) dominant disease allele to single and to flanking markers, with varying recombination fractions and numbers of marker alleles, and considering only affected individuals and parents when available. The simulated mean lod scores represent reasonable approximations of calculated expected lod (ELOD) scores. Also studied were sibships (two-four affected), effects of unavailability of one or both parents, and genotyping of additional well offspring to “reconstruct” the marker genotype of a deceased parent. These pedigree constellations provide significant linkage information, particularly with flanking markers at 5–11 cM intervals with PICs of at least 0.70, e.g. mean lod scores of 0.22–0.26 per informative meiosis (5 cM intervals, four equally frequent alleles per marker) for simple constellations. Pedigrees (with the same number of affected members) are more informative when the oldest available generation includes either a single ill progenitor (e.g. proband's parent or grandparent) or undiagnosed parents available for marker genotyping, rather than affected siblings (e.g. proband and a sibling, or proband's parent and uncle). On average, the availability of one parent for genotyping is almost as informative as two parents, and genotyping of two undiagnosed offspring can restore most of the information lost when one parent is unavailable. When parental marker genotypes are available, however, unaffected siblings provide little information when penetrance is estimated at 0.8 or less. An introduction to linkage information principles is provided, implications are discussed for prioritization of families for linkage studies, and relevance of the data for the study of more complex modes of inheritance is considered.