Agriculture Reference
In-Depth Information
former assumptions for the model we need to infer
Z
and
P
values, this is
done using a Bayesian approach.
Pr
Z
;
P
j
X
∼
Pr
Z
Pr
P
Pr
X
j
Z
;
P
:
It is not possible to calculate the given distribution exactly, but it is
possible to obtain an approximate sample (
Z
(1)
,
P
(1)
), (
Z
(2)
,
P
(2)
),...,
(
Z
(M)
,
P
(M)
) from Pr(
Z
,
P
X
) using Markov chain Monte Carlo (MCMC)
methods. To summarize, in structured association, individuals are
|
rst
allocated to populations, then population membership information is
used as a covariable in the test of association.
The software STRUCTURE is a good tool to
find population structure
when using the structured association approach. This software also
allows the researcher to estimate the proportion of ancestry attributable
to each population because sometimes individuals will belong not just to
one population; those individuals could be descendants of crosses
between two or more ancestral populations (Mackay and Powell 2006).
There is no way to determine which approach should be used for AM.
The selection of the approach should be made taking into account
sample strati
cation, the type of association (CG or WGA), the type of
phenotype, and the number of markers used. For this reason, we present
several approaches and for each one we present the considerations and
possible outputs.
IV. EXAMPLES OF ASSOCIATION MAPPING
Early on, most of the researches with AM approaches were done with
candidate genes, even though this is the AM approach that requires a
greater level of knowledge about the trait being studied. Recently, GWA
studies have become more frequent as marker abundance in most crops
has surpassed the critical threshold of thousands rather than hundreds of
markers. CG screens have generally been for the
with
widely known biochemical pathways or homologous genes from model
species, while GWA studies have been broader and tackled moderately
easy to dif
“
easy pickings
”
cult issues of phenotyping and genetics.
The CG study typically focuses on loci that are well characterized and
leaves no space for new genes to be discovered. The use of GWA,
meanwhile, evaluates random loci across the whole genome and can
serve as a kind of validation of the QTL mapping done before, but
requires LD estimates to be accurate. Furthermore, GWA requires LD
decay to be slow to moderate in the region of the QTL or a high density of
markers in those regions to capture the highest statistical association
