Peter L. Morrell Agronomy & Plant Genetics - Minnesota - SoyBase

39 Pages·2012·13.13 MB·English

Next Generation Crop
Improvement
Peter L. Morrell
Agronomy & Plant Genetics - Minnesota
Monday, February 27, 12
REVIEWS
Crop genomics: advances
and applications
Peter L. Morrell1, Edward S. Buckler2 and Jeffrey Ross-Ibarra3
Abstract | The completion of reference genome sequences for many important crops and
the ability to perform high-throughput resequencing are providing opportunities for
improving our understanding of the history of plant domestication and to accelerate
crop improvement. Crop plant comparative genomics is being transformed by these data
and a new generation of experimental and computational approaches. The future of
Morrell et al. 2012
crop improvement will be centred on comparisons of individual plant genomes, and
some of the best opportunities may lie in using combinations of new genetic mapping
Monday, February 27, 12
strategies and evolutionary analyses to direct and optimize the discovery and use of
genetic variation. Here we review such strategies and insights that are emerging.
The completion of reference genome sequences for many and Brachypodium distachyon7,8. Comparative genom-
Genome-wide association
important crops and model plants has the potential to aid ics — which is traditionally thought of as the analysis
studies
(GWASs). Studies that search in the realization of the long-standing promise of plant of synteny (gene order) and sequence comparisons
for a statistical association genomics to dramatically accelerate crop improvement1. among related species — is now being redefined by
between a phenotype and
Since the late 1960s, it has been possible to survey molec- the rapid publication of increasing numbers of refer-
a particular allele by screening
ular markers across a plant genome2, but for decades the ence genomes, by estimation of sequence diversity from
loci (most commonly by
genotyping SNPs) across number of markers that could be readily assayed placed high-throughput resequencing, by the examination of
the entire genome. limits on the genetic resolution that could be achieved the genomic distribution of large insertions and dele-
using either experimental or comparative genetic tions (indels) and copy number variants (CNVs) and
approaches. Only a few years ago, the highest-density by the emergence of a new generation of experimental
genetic maps required the laborious assay of several thou- and computational approaches. From genetic mapping
sand markers (for example, REF. 3). Experimental popula- to evolutionary analysis, the future of crop improve-
tions were generally limited to simple crosses between ment will revolve around the comparisons of individual
1Department of Agronomy
and Plant Genetics, two parents; more elaborate study designs that might plant genomes. Maximizing the use of this genomic data
University of Minnesota, provide an assessment of the genomic distribution of for crop improvement is of fundamental importance if
St Paul, Minnesota, 55108.
agronomically important mutations and their frequency we are to continue increasing crop production in the
2US Department of
in the relevant germplasm were proscribed by limits on face of growing human populations and changing cli-
Agriculture–Agriculture
marker technologies and the analytical approaches that mates while minimizing the environmental impact of
Research Service (USDA–ARS)
and Institute for Genomic could be used to distinguish the contribution of multiple agricultural activity.
Diversity and the Department
parents. Comparative approaches for the identification In this Review, we begin by addressing the chal-
of Plant Breeding and
of functionally important mutations based on analysis of lenges for comparative crop genomics that are posed by
Genetics, Cornell University,
marker frequency among populations had also been pro- the complex organization of plant genomes and the high
Ithaca, New York 14853, USA.
3Department of Plant posed4, but the high variance in expected allele frequency levels of nucleotide and structural diversity that are found
Sciences and the Genome between populations5 made the discovery of function- in many crop species. We then discuss the importance of
Center, University
ally important variants among the high number of loci understanding domestication, as the origin and demog-
of California Davis,
surveyed highly improbable. raphy of a crop affect the genetic basis of agronomic traits
California, 95616, USA.
A reference genome is now available for a number of and influence patterns of nucleotide diversity genome-
Correspondence to
P.L.M. and J.R.-I. crops (FIG. 1), and progress is being made towards refer- wide. We examine the ways in which our understanding
e-mail: [email protected]; ences for crops with large genomes6 (for example, see of the genetics of agronomic traits is being fundamentally
[email protected]
links in Further information). In addition, reference reshaped by genomic data. High-density genetic markers
doi:10.1038/nrg3097
genomes have been published for a number of other are being used in genome-wide association studies (GWASs)
Published online
29 December 2011 model plant systems, including Arabidopsis thaliana and can also be exploited for genomic selection.
NATURE REVIEWS | GENETICS VOLUME 13 | FEBRUARY 2012 | 85
© 2012 Macmillan Publishers Limited. All rights reserved
Soybean Genomics
Strategic Plan
•
Goal 1: Improve Utility of Genome
Sequence
•
Improve bioinformatics resources -
practical applications
•
Goal 2: Translational Genomics - Optimize
Breeding Efficiency
Boerma et al. 2011
Monday, February 27, 12
Topics
•
Next generation
populations
•
Selection against
deleterious mutations
•
Applications of genome-
wide SNP data
Monday, February 27, 12
Observational
Astronomy
•
Information about places we
won’t ever visit
•
The visible portion of the
electromagnetic spectrum is
only a fraction of what exists
•
DNA resequencing data latent
with information about the past
Monday, February 27, 12
DNA Resequencing
•
Most direct measure of
genetic diversity
•
Can assay all heritable
variation
•
Can now be collected
very rapidly
Monday, February 27, 12
Genome Size
R E V I E W S
18,000
No published sequence
16,000
Published sequence
14,000 Average angiosperm
12,000
)
b
M
e ( 10,000
z
i
s
e
m
8,000
o
n
e
G
6,000
4,000
2,000
0
Cucumber PeacSthrawberry Orange PapayaMedicFaogxotail millet Cacao Rice Grape CassavaSorghuPimgeonpea Potato TomatoSoybeaSnugar beet MaizSeugarcane BarBlreeyad wheat
Figure 1 | Crop genome size. Genome size of all published crop genomes (shown in green) and the five most
Nature Reviews | Genetics
important production crops with unpublished genome sequences (shown in blue). The average angiosperm genome
size of ~6 Gb is shown by the dotted line for comparison.
Understanding of agronomic traits is also being improved use of local patterns of linkage disequilibrium (LD), will
Monday, February 27, 12
by a new generation of multiparent genetic mapping pop- be useful for identifying paralogous reads in complex
ulations (or next-generation populations). As we discuss, crop genomes. Although there may be no simple solu-
higher-throughput resequencing and marker genotyping tion to the complexity of polyploid genomes, sequencing
will also enable new approaches towards crop improve- diploid relatives15,16 or double haploid lines17 can provide
ment, such as the identification and selective elimination a baseline for future genome-level research in polyploid
of deleterious mutations. crops.
The high levels of nucleotide diversity in some crop
Challenges of plant genomes genomes pose a challenge for comparative analyses, as
The genomic tools that are applied to plants are often higher numbers of mismatches between a sample and
developed for and tested against data from humans or a reference will result in reduced sequence read map-
other model systems, such as fruitflies or mice9,10, but ping (FIG. 2) or reduced hybridization to oligonucleotide
the size and dynamic nature of plant genomes adds to arrays. For example, the maize and human genomes are
or exacerbates challenges that are faced in other systems similar in size, but an average pair of maize individuals
(FIG. 1). Plants tend to have a larger number of multi- differs at tenfold more sites than any two humans
gene families11 and a higher frequency of polyploidy do18. Although many crops do not have high levels of
than occurs in mammals. This makes paralogy a more diversity, the difficulties of a diverse genome are not
substantive issue because the short sequence reads unique to maize as an outcrossing species: diversity is
that are typical of high-throughput sequencing may also high in the clonally propagated grape19 and even in
Paralogy not map uniquely to a reference genome, and allelic self-fertilizing (‘selfing’) species, such as barley20.
Unlike orthologous genes,
variation cannot then be distinguished from differences Another challenge in plant comparative genomics is
which trace their common
among closely related gene family members (FIG. 2). genome size (FIG. 1). Plant genome size varies by more
origin to a locus in an ancestral
Paralogy remains a problem even in plant species that than three orders of magnitude in currently character-
species, paralogous loci consist
of gene copies that trace their have a high-quality reference genome owing to the ized species21, largely owing to the prevalence of trans-
common origin to a duplication prevalence of extensive copy number variation12,13. For posable elements22. Size alone makes genomic analysis
event within a genome.
instance, estimates suggest that the maize reference more difficult: shotgun sequencing reads that are suf-
genome accounts for only ~70% of the low-copy-number ficient to provide deep (25×) coverage of four Drosophila
Linkage disequilibrium
sequences that are present in the parents of a diverse set melanogaster genomes — enabling the identification of
(LD). Nonrandom association
of alleles at two or more loci. of maize inbreds and that this copy number variation heterozygous sites and structural variation — would pro-
The pattern and extent of LD leads to a high percentage of false-positive variants14. It vide a meagre ~1× coverage of the wheat genome. The
in a genomic region is affected
seems likely that continued improvement in sequence density of transposable elements in plant genomes also
by mutation, recombination,
read length, along with methodological approaches that means that a large fraction of shotgun sequencing data
genetic drift, natural selection
and demographic history. assess allelic segregation among lines14 and that make is of limited use for reference-based genomic analysis, as
86 | FEBRUARY 2012 | VOLUME 13 www.nature.com/reviews/genetics
© 2012 Macmillan Publishers Limited. All rights reserved
Short Read Mapping
R E V I E W S
a
Demographic history and geographic origins.
b
Genome-wide polymorphisms make it possible to
c
examine the demographic history and geographic
d origins of crops. Domestication is an evolutionarily
e ? recent phenomenon, and most of the genealogical
history at any locus will be shared between a domesti-
cate and its wild progenitor28. Comparisons of alleles
within and between domesticated and wild taxa will
reveal divergence times that greatly predate the origin
of the cultivated form29,30, reflecting the time to most
Transposable elements Genes Gene with SNP
recent common ancestor of the species rather than
the time of divergence of the domesticate. A detailed
Figure 2 | Challenges of read mapping in plant genomes. The mapping of short understanding of domestication history requires a
Nature Reviews | Genetics
sequence reads to a reference plant genome is shown with the genome at the bottom large number of loci in conjunction with modelling
and with sequencing reads above. Coloured shapes represent transposable elements
of population demography. Some of the earliest work
or genes; the two orange ovals represent a pair of paralogous genes. Short sequence
on demographic modelling in plants used mean pat-
reads are shown directly above where they would map to the reference. Different
terns of genetic diversity to fit a bottleneck model of
Morrell et al. 2012
scenarios are shown in lines a–e. a | Uniquely mapping reads, including junctions
domestication31, an approach that was later extended
between sequence repeats. b | A sequence from a diverse genome that would fail to
Monday, February 27, 12 to include an explicit likelihood framework32,33. More
map to the reference owing to an excess of SNP differences. c | A read from one
recently, investigators have used methods that incor-
paralogue that maps incorrectly owing to a sequence error or a SNP. The correct
porate more detailed information, such as the site
mapping is shown with a grey read. d | Reads that would map multiply and are usually
filtered from further analysis. e | A read from a third copy of the orange gene that is frequency spectrum34,35, to distinguish among different
incorrectly mapped to one of the reference copies, leading to a false SNP. This is likely evolutionary models.
to be the result of a copy number variant that was not included in the reference One of the most fundamental issues that influ-
genome (as indicated by the question mark).
ences the genetic architecture of agronomic traits and
the levels of genetic diversity in crop genomes is the
number of times that a species has been domesticated.
reads map with equal probability to multiple positions There are compelling examples for both single domes-
in the reference (FIG. 2). It is not surprising that the crop tications (such as maize and soybeans)36,37 and multiple
genomes that have been sequenced to date have all been domestications (such as avocados, common beans and
relatively small — the largest crop genome sequenced, barley)38–40, but the number and location of domestica-
maize, is less than half the size of the average angiosperm tion events for most crops remain unresolved. Simple
genome (FIG. 1; TABLE 1). statistical methods that cluster individuals or popula-
Although plant genomes pose a number of challenges tions based on genetic diversity within the domesticate
Bottleneck
for genomic analysis, they do offer some advantages. can be misleading, as the number of genetic groupings
A temporary marked reduction
Unlike most animals, crops can be propagated clonally is not necessarily reflective of domestication history41,42.
in population size.
or maintained as inbred lines, and the seeds of many spe- For example, although genetic evidence suggests two
cies can be stored indefinitely, which effectively immor- domestications of the common bean39, genetic drift in
Site frequency spectrum
The distribution of allele talizes genotypes of interest. This makes it possible to cultivated populations leads to the identification of
frequencies in a population:
sequence a line once but to phenotype the line many multiple genetic groups43.
essentially a count of the
times, and it allows replication across environments23. The details of even the simplest of domestication sce-
number of alleles in a
Inbred lines or specially created double haploids also narios are likely to be complex. For example, geograph-
population at a given
frequency. avoid the difficulties of sequencing highly heterozygous ical spread of the domesticate followed by admixture
genomes. Sequencing of the grape genome has provided with wild relatives can obscure geographic origins44,45.
Genetic drift
a useful comparison of the advantages and difficulties Extensive admixture may be one explanation for
Fluctuations in allele
of sequencing a diploid outcrossing accession13 or an the continued controversy regarding the origins of
frequencies that are due to the
inbred line24. the domesticated indica and japonica subspecies of rice.
effects of random sampling.
Analyses from recent genome-wide resequencing have
Admixture Origin and evolution of crops
failed to reach a consensus on the number of domesti-
The mixing of two or more
Understanding the origins and domestication of crop cations of rice: modelling of genetic differentiation sup-
genetically differentiated
plants is of substantial evolutionary interest, as domesti- ports separate domestications followed by introgression
populations.
cated plants provide a model system for studying adapta- at agronomically important loci46, whereas the site
Introgression tion25,26. An understanding of crop origins has long been frequency spectrum and phylogenetic analysis of mul-
The incorporation of genetic
held as central to the identification of useful genetic tiple data sets argue for a single origin35. As whole-
material from one population
resources for crop improvement27. Domestication shapes genome data become available for more crops and
or species into another by
the genetic variation that is available to modern breed- their wild relatives, application of methods that make
hybridization and backcrossing.
ers as it influences levels of nucleotide diversity and better use of additional information from detailed
Haplotype
patterns of LD genome-wide. The demographic his- haplotype structure and patterns of admixture across
The combination of alleles
tory of domestication also informs our expectations of the genome (for example, REFS 47,48) will improve
or genetic markers found
the genetic architecture of traits and thus our ability to insight into the complex demographic histories of
on a single chromosome
of an individual. identify causal genetic variants for crop improvement. many crops.
NATURE REVIEWS | GENETICS VOLUME 13 | FEBRUARY 2012 | 87
© 2012 Macmillan Publishers Limited. All rights reserved
Molecular Population
Genetics
resequencing
human Hap
in Drosophila
Map (1M SNPs)
(Kreitman)
2nd generation
GenBank
human Hap Map
begins
(3.1M SNPs)
allozyme DNA
diversity in sequencing resequencing maize genome
Drosophila (Maxam & in maize sequenced
(Lewontin & Gilbert, Sanger (Shattuck- human genome
Solexa DNA
Hubby) et al.) Eidens et al.) sequenced
sequencing
1960 1970 1980 1990 2000 2010
infinite alleles estimator of estimator of estimators forward time
model θ = 4Neµ ρ = 4Ner of ρ simulation
(Kimura & (Watterson) (Hudson) (Hudson) (Nordborg,
Crow) (Kuhner, Thornton, et
coalescent
Yamato, al.)
infinite sites model theory
Felsenstein)
(Kimura) (Kingman)
(Wall)
4 gamete
neutral theory (Fearnhead
test
(Kimura) & Donnelly)
(Hudson &
Kaplan)
neutral theory Approximate
(Kings & Jukes) coalescent Bayesian
with Computation
recombination (Pritchard et
(Hudson) al.)
Monday, February 27, 12
Derived Site Frequency
Spectrum (SFS)
GGGATGGCC......GGCACGGGC
.......................G
..C...A.............C..G
........................
..C...A..........G......
..C...A................G
........G...............
..C...A.............C..G
4 4 1 1 2 4
Monday, February 27, 12