Agriculture Reference
In-Depth Information
approach will rapidly become obsolete as sequencing costs are dropping
to the point where whole-genome sequencing (WGS) (Deschamps and
Campbell 2010) or genotyping by sequencing (GBS) (Elshire et al. 2011)
will be more cost-effective.
Improvements in genotyping will also require error rates to decline as
this is relevant for association genetics analysis because low error rates
(
3%) have large effects on the accuracy of LD estimates, affecting the
accuracy of association especially for low-frequency alleles. Random
selection of SNP markers is important for LD estimation. Nonrandom
samples of SNPs result in ascertainment bias, which undersample low-
frequency mutations. As a result, a selection of SNPs occurring at
intermediate frequencies is a more optimum strategy, reducing the
estimates of LD compared with those made from random SNP evaluation
(Ingvarsson and Street 2010).
The number of markers needed for AM depends on the type of
approach chosen (CG or GWA) and also relies on an estimate of LD
decay. In turn, LD decay depends on the mating pattern of the species
selected. This will determine the amount of markers needed in order to
have good resolution in the association study. For inbreeding species in
which LD decay is slow and extended through several kilobases (Kb) or
even megabases (Mb) of sequence, there is a need for many fewer
markers than in outcrossing populations that have less extended LD.
This is a key component to decide what number of markers is required
for any type of AM. For example, while 140,000 markers provide a
reasonable coverage of the 125Mb
Arabidopsis
genome, a rough esti-
mate suggests that over a 2 million markers will be required to cover
the 475 Mb genome of grape, and 10
<
15 million may be necessary for
diverse maize cultivars (Myles et al. 2009). It is also important to take
into account marker locations, because in order to do a cost-effective
genotyping, fewer markers are needed in centromeric regions and
more markers should be from regions of high recombination and gene
concentration.
Candidate gene analysis assumes knowledge of the genetics of the
character, with extensive use of annotated databases being important to
-
find genes related to the trait under study. Sequencing candidate genes
offers a lot of possibilities for SNP discovery and can help in the
discovery of different haplotypes for the association analysis. Unlike
for GWA, SNPs as well as other markers can be used in candidate gene
analysis. However, next-generation sequencing will make it possible for
future projects to cost-effectively obtain full sequence data from large
population samples, making bioinformatics evaluations of candidate
genes possible.
