Summary: exclusion mapping Arguments: < relative risk ratio hypotheses >
Usage --> command line:
npl:6> exclude
You are then given the option of inputting a set of z's or relative
risk values (the input queries will be different depending on whether
you want to analyze your data under the assumption of no dominance
variance or not).
Output --> The text file consists of tabbed columns in the format:
position z2-1 z2-2 z2-3 ... etc.
(You should be able to use this file as input to a plotting program if
you don't have access to a postscript printer.) At the end of the
text file is a time-stamped summary of the session settings. The
postscript file consists of multiple y-axis LOD score plots for each
relative risk value/set of z's and gives the distance examined in the
analysis at end of the x-axis (this may be larger than the map
distance if you have specified an off-end distance). A horizontal
dashed line is drawn at the traditional exclusion criterion of Z < -2.
Background --> Exclusion mapping is used to identify and exclude regions unlikely to have a major effect on the trait you are mapping. GENEHUNTER does this by comparing the likelihood of the observed sharing proportion of 0, 1 and 2 alleles between affected sibs (z0,z1,z2), to the likelihood under the Mendelian expectation of a0=1/4, a1=1/2, and a2=1/4. When using GENEHUNTER under the assumption of no dominance the sharing proportions are given by:
z0 = a0/Ls
z1 = a1
z2 = a2((2Ls-1)/Ls)
where Ls = lambda-sub-S, the relative risk ratio for a sib, defined as:
prevalence of the trait in siblings of affected individuals
---------------------------------------------------------
prevalence of the trait in the population at large
Note that Ls = 1 when there is no observed difference in prevalence of
sibs vs the population (z0=a0, z1=a1, z2=a2 and LOD = 0). If Ls < 1,
it would imply that there was some protective advantage in having an
affected sib. Since neither of these cases are interesting and/or
reasonable, only Ls values > 1 are allowed. (The no dominance
variance assumption allows us to simplify the sharing proportions
above to the one variable Ls. With dominance variance Ls = Lo where
Lo = relative risk ratio for an offspring, and Lm-1=2(Ls-1) where Lm
is the relative risk ratio for a monozygotic twin.)
The likelihood under Bayes theorem is:
L(pos) = (z0*p0+z1*p1+z2*p2) / (a0*p0+a1*p1+a2*p2)
and the LOD score is calculated by summing log10(L(pos)) across
pedigrees for each position.
The relations for z0, z1, and z2 above hold if multiple loci are involved in the trait, provided that the loci interact multiplicatively and the lambda values are defined as the component of the relative risk attributable to the locus.
More details on the analytical method are present in the publication...