Behavior Genes
601
can take values between 0 and 1. How-
ever, in order for linkage studies to be
carried out accurately, the penetrance pa-
rameters together with the disease gene
frequency need to be speciFed correctly.
Otherwise, at best, the estimate of recom-
bination will be incorrect, and at worst,
linkage will not be detected even when it
is present.
Other problems include possible genetic
heterogeneity in common disorders and
relative imprecision in deFning who is
affected or unaffected within families.
This is not a difFculty when there are
clear-cut cases and clear-cut healthy family
members, but common diseases often also
show milder or uncertain cases within the
same families. The most straightforward
way of dealing with these problems has
been to concentrate on extended families
with many affected members and to
make an informed guess at the mode of
transmission. It is also assumed that even
if there is heterogeneity in the disorder
as a whole, there is homogeneity at
least within the families. Such simplifying
assumptions have worked well with some
common diseases, with the most notable
success
amongst
psychiatric
disorders
being
early-onset
familial
Alzheimer’s
disease. However, in other disorders such
as schizophrenia, concentration on large
multiply affected families has produced
results that are more difFcult to interpret,
hence there has been a move toward using
nonparametric or ‘‘model-free’’ methods
of analysis.
Nonparametric linkage analysis
The most
common approach concentrates on af-
fected sibling pairs. Affected sib pairs may
be more representative of a common dis-
order than subjects found in multiplex
pedigrees and, in general, Fnding fami-
lies containing affected sib pairs is easier
than Fnding extending pedigrees. How-
ever, the real advantage of focusing on sib
pairs who are both affected by the disorder
is that the statistical analysis is relatively
straightforward and robust. Detection of
linkage depends on the fact that at any
given locus, the probabilities of a pair of
siblings inheriting 0, 1, or 2 alleles that are
identical by descent are respectively 1/4,
1/2, and 1/4. Greater allele sharing at a
marker than would be expected by chance
in pairs of siblings who are both affected
by the disease suggests that the marker
locus is close to a gene conferring sus-
ceptibility to the disorder. The unfortunate
disadvantage of sib pair methods is that
although they are simple and robust, there
is a comparative lack of statistical power.
Both practical experience and simulation
studies suggest that unless the sample
sizes are extremely large, linkage is dif-
Fcult to detect with genes that confer a
relative risk of a disorder of much less
than about three or account for less than
about 10% of a variation in liability to
the disorder. Genes with effects as large
as that may be rare in common diseases
and so, for example, a genome scan for
linkage in schizophrenia in almost 200
families effectively excluded any gene hav-
ing a relative risk of three from most of
the genome. The need to detect genes hav-
ing only small effects has brought about a
renewed interest in applying allelic associ-
ation studies in a more systematic fashion
than just focusing on potential candidate
genes.
3.2.2
Association Studies
Allelic association is the phenomenon
whereby a particular allele at a marker
locus is found together with a trait or
disorder more than would be expected by
chance. In contrast with linkage, the same
allele is found across the population rather
previous page 601 Encyclopedia of Molecular Cell Biology and Molecular Medicine read online next page 603 Encyclopedia of Molecular Cell Biology and Molecular Medicine read online Home Toggle text on/off