6 Identity by Descent in the Mapping of Genetic Traits
155
practical only for individuals with fully observed genotypic data. Second, there
are multiple individuals who may be related outside the defined pedigree structures. Population-based realizations of IBD jointly among multiple individuals are
intractable.
One case where analysis is tractable is that of a set of parent-offspring trios for
whom there are unphased genotype data. Population-based modeling of the IBD
between the two parents can be combined with the modeling of the two meioses
from parents to offspring. Joint analysis of the six genomes of the trio is important:
pairwise analyses are much less informative and can give conflicting results. With
joint analysis, one may, for example, compare the location-specific IBD probability
between the two genomes of the offspring (i.e., autozygosity) with that expected
given the IBD probabilities between the parental genomes. Hence one may detect
regions of the genome in which autozygosity of surviving offspring is significantly
lower than expected given the inferred segments of IBD in the parents, indicating
possible selection against autozygosity in these regions.
In more general pedigrees, there is an additional complication in combining
population- and pedigree-based IBD estimates. Within a pedigree, the maternal/paternal origins of haplotypes can be realized where there are informative
data. In other cases, for example, for the two haplotypes of founders, there is no
information (even with data) on which haplotype is maternal and which paternal, but
this is irrelevant to within-pedigree IBD. In the population-based context, even if the
IBD inference implies fully correct phasing, it is likewise arbitrary which haplotype
is designated the maternal/paternal one of the individual. Combining population and
pedigree IBD faces the intractable challenge of resolving the multiple pairings of
each individual’s two haploid genomes with the labeling in each population-based
realization of IBD. Thus while IBD provides a natural unifying framework in which
to combine pedigree- and population-based inferences, there remain challenges for
successful implementation of methods.
6.5
Summary
We have shown how IBD Z can be inferred from genetic marker data X and then
used to provide evidence for genome locations at which the DNA variants may
be causal for trait phenotypes Y. Rather than considering directly the association
between X and Y, we consider this association through the lens of descent Z. In
fact, a basic assumption of our models is that X and Y are conditionally independent
given Z. In the three main sections of the text, we have considered: first, probability
models for Z; second, inference of Z fromX; and third, use of this inferred or
realized Z to map DNA underlying Y
In Sect. 6.2 the focus was on the models for IBD or Z. We considered probability
models for Z, both in defined pedigrees and among members of a population. It is
important to consider not only IBD at separate locations but also how it changes
across a chromosome. Because DNA descends generation to generation is large
segments, even remote relatives will (if they share any genome IBD) share segments
155
practical only for individuals with fully observed genotypic data. Second, there
are multiple individuals who may be related outside the defined pedigree structures. Population-based realizations of IBD jointly among multiple individuals are
intractable.
One case where analysis is tractable is that of a set of parent-offspring trios for
whom there are unphased genotype data. Population-based modeling of the IBD
between the two parents can be combined with the modeling of the two meioses
from parents to offspring. Joint analysis of the six genomes of the trio is important:
pairwise analyses are much less informative and can give conflicting results. With
joint analysis, one may, for example, compare the location-specific IBD probability
between the two genomes of the offspring (i.e., autozygosity) with that expected
given the IBD probabilities between the parental genomes. Hence one may detect
regions of the genome in which autozygosity of surviving offspring is significantly
lower than expected given the inferred segments of IBD in the parents, indicating
possible selection against autozygosity in these regions.
In more general pedigrees, there is an additional complication in combining
population- and pedigree-based IBD estimates. Within a pedigree, the maternal/paternal origins of haplotypes can be realized where there are informative
data. In other cases, for example, for the two haplotypes of founders, there is no
information (even with data) on which haplotype is maternal and which paternal, but
this is irrelevant to within-pedigree IBD. In the population-based context, even if the
IBD inference implies fully correct phasing, it is likewise arbitrary which haplotype
is designated the maternal/paternal one of the individual. Combining population and
pedigree IBD faces the intractable challenge of resolving the multiple pairings of
each individual’s two haploid genomes with the labeling in each population-based
realization of IBD. Thus while IBD provides a natural unifying framework in which
to combine pedigree- and population-based inferences, there remain challenges for
successful implementation of methods.
6.5
Summary
We have shown how IBD Z can be inferred from genetic marker data X and then
used to provide evidence for genome locations at which the DNA variants may
be causal for trait phenotypes Y. Rather than considering directly the association
between X and Y, we consider this association through the lens of descent Z. In
fact, a basic assumption of our models is that X and Y are conditionally independent
given Z. In the three main sections of the text, we have considered: first, probability
models for Z; second, inference of Z fromX; and third, use of this inferred or
realized Z to map DNA underlying Y
In Sect. 6.2 the focus was on the models for IBD or Z. We considered probability
models for Z, both in defined pedigrees and among members of a population. It is
important to consider not only IBD at separate locations but also how it changes
across a chromosome. Because DNA descends generation to generation is large
segments, even remote relatives will (if they share any genome IBD) share segments