MLINK: impossible result?
Mr J. Attwood
jattwood at hgmp.mrc.ac.uk
Tue Aug 22 07:16:56 EST 1995
In article 63962 at ucl.ac.uk, dcurtis at hgmp.mrc.ac.uk (David Curtis) writes:
> Because the disease allele has a frequency of 0.001, there is a small
> though not infinitesimal probability for there to be two different
> disease alleles entering the pedigree, one segregating in one branch
> and another segregating in the other branch. Thus your results are
> just what I would expect.
Yes, I agree with Dave. You could try giving individual #2 a normal parent, in
order to see whether you do, then, get -infinity at a theta of zero. This will tell
you whether the program is considering the possibility that a founder was homozygous
for the disease allele. I haven't drawn out the pedigree, but I suspect that this is
the most likely case, as you don't have anyone whose disease status is unkown and
therefore it is not possible for two unrelated members to be independently bringing a
disease allele into different branches of the pedigree.
I think that it is a general rule that, when the Linkage package comes
up with surprising results, there is usually a second possibility which it is taking
into account and whose lodscore curve is modifying the expected one. The strength of
the Linkage package is its ability to fill in for missing data, and it often allows
for things, like this, not immediately obvious to humans.
More information about the Gen-link