ALTERNATIVE TEST FOR LINKAGE BETWEEN 2 LOCI

When using the maximum likelihood method to test for linkage it is tra ditionally assumed that the female (theta(f)) and male (theta(m)) reco mbination fractions are equal. However, this assumption is not always realistic. In this paper we present a test for linkage that does not r equire this assumption. Specifically, we propose testing for linkage b y testing the null hypothesis H-0: theta(f) + theta(m) = 1 vs. the alt ernative H-A: theta(f) + theta(m) < 1, treating theta(f) - theta(m) as a nuisance parameter. This leads to a likelihood ratio test statistic that is asymptotically distributed as a chi-square with one degree of freedom. By examining the expected values of the maximum lod scores, we show that for data from phase-known meioses this test can provide a more powerful test for linkage-especially so when theta(f) not equal theta(m)-unless the recombination fraction is zero. For data from phas e-unknown meioses, the proposed test is a more powerful test for linka ge only for large values of \theta(f)-theta(m)\; otherwise, the tradit ional test has higher power. Thus, the proposed test can lead to gain in power for detecting linkage when either phase is known for most mei oses or when there is a large absolute difference between the male and female recombination fractions. (C) 1997 Wiley-Liss.