Genomics Assisted Plant Breeding Flashcards
What are landraces?
Lines that have been developed by many generations of placing seed in field and adapting it slowly. Halfway between wild and cultivar, lots of diversity, locally adapted.
Primitive varieties that have been cultivated over centuries or even millennia.
Tend to be locally adapted to survive in the region where they are grown.
Have low but stable levels of productivity.
Contain high diversity of potentially useful genetic variation but far fewer gene alleles that are deleterious for agriculture than wild lines.
Also have some less desirable properties than cultivars, but always produce good yields.
What is association mapping?
Each circle is a plant.
Each letter is a genotype/marker score.
Each colour is a trait.
Looks at association between genotype and trait in population.
Gene is sitting reasonably close to the trait in the population.
Every sample is scored for thousands of markers to give dense coverage across the genome.
Need markers everywhere, otherwise won’t pick up associations for every gene.
And every line is scored for multiple traits in replicated field trials
What is Genome Wide Association (GWA) mapping?
All of the phenotype data are compared with all of the genotype data and the output is a huge set of association values linking every marker with its corresponding effect on the total phenotype.
(Each trace is relative to one phenotype.
All 7 chromosomes laid end to end.
Like QTL analysis.
Each peak is an association between a gene and a property.
Peaks are sharp- confidence intervals are small.
Properties are related eg. Plant more likely to fall over if it is taller.
Breaking at neck tends to occur if there is a lot of seed and it is heavy.)
Why is association mapping more accurate than ordinary QTL mapping?
Accuracy is proportional to the density of recombination events on the chromosomes.
Standard QTL mapping is derived from 2 generations of crossing between 2 parents – if you collect 100 recombinant inbred lines the total number of recombinations per chromosome is ~ 1.5 x 100. (QTL uses bi parental populations- correspondence between traits and markers dependent on number of recombination events.)
Association mapping uses a collection of lines that are separated from each other by anything up to 100 years of crossing – so there are far more recombination break-points in the population. (Association uses hundreds of lines only related by ancient history (not pedigree)- could be separated by hundreds of generations- many recombinations separate them.)
What do recombination hotspots do?
Recombination hotspots disrupt even distribution of breakpoints.
‘Linkage disequilibrium’ (LD) is essentially equivalent to the density of breakpoints.
Recombinations rates vary dramatically across chromosomes.
Some areas are prone to recombination, some are not- both can skew results.
High amounts of previous recombination – linkage disequilibrium.
Essentially same as no. breakpoints.
High LD= lots of breakpoints.
How are alleles linked in cultivars, landraces and wild barley?
Virtually every allele is linked to every other allele in cultivar. Cultivar association mapping- will get a peak, but don’t know where it is cause everything is so closely linked. Only need a small population, but peaks will be back into areaa of QTL.
Landraces are halfway- patchy regions.
Wild- so much recombination, nothing linked to anything else (or very little).
The wilder the collection gets the lower the LD becomes.
What is the optimal gemplasm collection for GWA mapping?
Genetically unstructured- not grouped into a tree.
In unstructured population, only cultivars group together. Almost no substructure in the rest of the population.
What is the “wild germplasm” problem?
If you cross a cultivar with a wild line then follow the Pedigree Breeding strategy, the only traits you can see improvement in are biotic resistances and abiotic tolerances because the many disadvantageous alleles mask the other traits. (Can’t score QTL in progeny of wild x cultivar- traits are masked.)
How can the wild germplasm problem be solved?
One way to do this is to construct a Nested Association Mapping population.
Filtering amount of wild germplasm you have in population- reducing so genotype is scoreable.
Every line is mostly cultivar, but pieces are derived from one of multiple wild lines.
Have over 1000 lines.
Never get different colours in a line- each line only has two parents- cultivar and one known wild line.
Backcross strategy.
Each number actually represents up to 50 lines.
The other 49 in each group will have a different pattern of wild type lines.
Each line is about 20% wild plant and 80% cultivated- can score traits in these.
Get different patterns of germplasm in each family.
Each Family derives from 2 parents – a wild line and the common cultivar used for all lines. Thus, each Family has its own individual collection of segregating traits.
Need more than 60 lines to reliably detect a QTL- can’t just use one line (in statistical terms).
If there is a useful gene at a certain location in one line, likely that there are useful genes in other lines- can combine data to get good results.
QTL cannot be reliably detected within 60-member families, But many QTL will be segregating within multiple families and they can then be detected.
