Linkage Strategies for Genetically Complex Traits.
III. The Effect of Marker Polymorphism on Analysis of Affected Relative
Pairs
N. Risch
American Journal of Human Genetics, 46, 242-253 (1990)
[published erratum appears in
American Journal of Human Genetics, 51(3), 673-675 (Sep 1992)]
Abstract
The results from the second paper of this series are reexamined for markers that
are not completely polymorphic. A maximum lod score (MLS) criterion is defined for
affected relative pairs. The expected MLS (EMLS) is calculated as a function of the marker
polymorphic information content (PIC) for various values of lambda R (relative risk ratio)
and different relative types by using simulations. An m-allele model with equal allele
frequencies is employed. The EMLS is calculated for two sampling strategies: scheme 1,
which uses pairs only, and scheme 2, which also includes additional informative relatives.
For scheme 2, the percent of the maximum achievable EMLS (i.e., for a marker with a PIC
of 1.0) is approximately equal to the marker PIC value for all relative types. For scheme 1,
the EMLS is greatly diminished unless PIC is high, especially for distant relatives. For
example, scheme 1 is not cost-effective for sibs unless PIC greater than .7; for second- and
third-degree relatives, PIC must be greater than .85. Therefore, in general, it will be
worthwhile to type additional relatives in linkage studies using affected pairs. The
comparative value of sibs versus distant relatives depends on lambda R, recombination
theta, and PIC. For large lambda R and PIC values, distant relatives are preferred.
Alternatively, for smaller lambda R and PIC values, sibs are best.
Allele-sharing linkage analysis