A likelihood ratio statistic is proposed for combining two-point genet
ic linkage analyses when the two-point analyses are between a trait an
d a well-defined map of markers. It is assumed that the two-point anal
yses are independent, as in the case of choosing only the most informa
tive marker per family. The asymptotic distribution of the likelihood
ratio statistic is derived under the null hypothesis of no linkage of
the trait with a map of 2 markers, with intermarker genetic distance D
elta. This distribution is shown to be a chi-square mixture distributi
on with mixing probability depending on Delta and the assumed mapping
function. We use this asymptotic result to approximate the distributio
n of the likelihood ratio statistic for the more general case of more
than 2 markers. Simulation results indicate that this may be reasonabl
e. Power is evaluated by simulations and results indicate that this ap
proach, which constrains the intermarker distances to their known valu
es, tends to be more powerful than other methods proposed in the liter
ature. (C) 1994 Wiley-Liss,Inc.