Locating human quantitative trait loci: guidelines for the selection
of sibling pairs for genotyping
L. Eaves, J. Meyer
Behavior Genetics, 24(5), 443-455
(Sept 1994)
Abstract
Simulation studies were conducted to assess the relative merits of
different nonrandom sampling strategies for the selection of sibling
pairs for genotyping in the attempt to locate individual loci (QTLs)
contributing to variation in human quantitative traits. For a
constant amount of variation contributed by a QTL (25% of the total)
the frequencies and dominance relationships of a trait increasing
allele were varied. Three strategies for selection of pairs for
genotyping were based on the phenotypic values of the siblings:
"Concordant sib pairs" (CSP) are pairs in which both individuals
exceed a given threshold value; "discordant sib pairs" (DSP) are
pairs in which one member exceeds a given upper threshold and the
other is below a specified lower threshold; and "most similar pairs"
(MSP) are pairs selected for falling below a specified percentile
ranking of the within-pair mean square for the quantitative trait.
Tests for linkage with markers at 1, 2, 5, 10, and 20 cM from each
of the QTLs were conducted for each of the selected samples and
compared with tests based on the regression, in the entire sample,
of within pair variation on the proportion of alleles identical by
descent (IBD) at each marker locus. Tests for the effect of the
increasing allele at the QTL ("candidate gene") were also conducted
for the DSP pairs. No single nonrandom selection procedure yields as
much as half the information realized in the total sample. However,
a combined strategy which involves genotyping the 5% of MSP and DSP
for the upper and lower quintiles of values of the quantitative
trait (a further 3% of the sample approximately) yields lod scores
which are usually more than 65% of the values realized for the
entire sample. Tests comparing the proportion of increasing alleles
in high- and low-scoring siblings from DSP samples are uniformly
very powerful for detecting candidate loci. Even when it is not
possible to measure the entire range of the phenotype with uniform
precision, some attempt to differentiate among individuals in a
common "unaffected" class of individuals can lead to considerable
increase in power.
Allele-sharing linkage analysis