Biology Reference
In-Depth Information
MORE COMPLEX MODELS: HOST/WORM GENETIC
EFFECTS AND SPATIAL FACTORS INFLUENCING
HOST
ASCARIS
BURDEN
We have so far only considered simple models of host genetic varia-
tion influencing observed Ascaris burden phenotypes in humans.
However, greater model complexity may be required. Genetic variation
in the worms (see Chapter 8) is likely to influence these phenotypes, as is
shared environmental determinants. Using our variance component
framework, it is possible to test more complicated (and biologically
reasonable) models regarding the potential determinants of host worm
burden. We have recently published the first simultaneous examination
of host and worm genomic effects on host burden phenotypes
64
that we
summarize here. We initially wanted to determine if both host genes and
worm genes are important for determining the distribution of Ascaris
worm burden in the Jirel population. For this analysis we utilized data
available from 320 individuals (including 141 males and 179 females
ranging in age from 3 years to 79 years) for whom we had collected all of
the worms expelled post-albendazole treatment. For this analysis, our
measure of worm burden was direct count of worms obtained over
a 4-day fecal collection period. The 1094 Ascaris worms collected from
these individuals were then characterized for a total of 23 genetic
markers.
65,66
Using variance decomposition theory, we derived a model that allowed
the infection-related phenotypic variance to be decomposed into
components that are due to host genetic factors, genetic variation among
the worms, and random environmental factors. Expanding upon our
previous quantitative genetic model (Eq.
(12.2)
), the resulting phenotypic
covariance matrix is then given by:
U ¼ð2Fh
2
þ R
w
w
2
þ Ie
2
Þs
p
(12.4)
R
w
represents the estimated coefficient of relationship
67
between para-
sites across hosts (and averaged within host) as measured (see
65,66
) from
highly polymorphic short tandem repeat polymorphism in the Ascaris
worms collected from the Jirel human host population, and
is the
identity matrix. The additive genetic heritability in the worms is denoted
by w,
2
and the proportion of phenotypic variance due to random envi-
ronmental factors is now e
2
I
w.
2
Using this approach, we tested whether nested models that eliminate
specific variance components adequately captured the variation accoun-
ted for in this general model. Likelihood ratio tests (LRT) were formed as
twice the difference in ln likelihoods. See Williams-Blangero et al.
64
h
2
¼
1
for
technical details incurred by testing of boundary conditions.
