Introductory statistical genetics for plant breeding 1

Published on April 27, 2016   58 min

Other Talks in the Series: Statistical Genetics

0:00
I'm Fred van Eeuwijk, and I'm working at the Statistics Department of Wageningen University in the Netherlands. And I will tell you today about statistical genetical principles that are useful for plant breeders.
0:14
An overview of topics that we'll be dealing with today, at first I have a few slides as a general introduction to the topic, then I will talk about what I will call genotype to phenotype models, which we will see formed as the major tool in plant breeding. And important parts, lets say, of genotype by phenotype models is actually the part of the deals of the so called the genotype by environment interaction. And we will be talking a little bit about that. Then, of course, in plant breeding, we use phenotypic data for our modeling. And I will say a little bit about how we get those phenotypic data, which experimental designs we use, and how we elaborate those into statistical models. Then, I will say a few things about causal and observational parameters in statistics, while we estimate certain parameters using our statistical models. But in genetics, we use sometimes slightly different parameters that are based on our IDs about gene transmission, how alleles are transmitted through pedigrees from parents to off spring. Then, I will summarize what we have done, that's Intermezzo that you see. We continue in deeper view on what genotype to phenotype models are, and then we get two examples of these genotype to phenotype models. We will talk about QTL mapping, QTL mapping strategies as an example of how to you use genotype to phenotype models. We will look at an analysis of genotype by environment interaction by a model that is called Finlay Wilkinson model, Finlay Wilkinson regression. And then we will finish with an example of what we call multi-environment trial QTL mapping in which we use a rather advanced statistical model to find out which genes and QTLs are driving our phenotypic variation.
1:54
We are now in our introduction, and we first have some thoughts about, well, what are the objectives in plant breeding really? A major objective is to improve the phenotypic trait values of our plants, of our cultivars by actually improving their genetic properties. And that's the first step. If, let's say, we have identified the important genes, QTLs in usually different individuals, we have to compile these genes, these alleles in such a way that let's say, we get them together individual as a kind of Idiotype, an ideal desirable genotype, the strategy of compiling those alleles, those genes is the second major objective in plant breeding. We will talk mainly about, let's say, the first objective. So the first objective actually means that we need to identify the genes or the Quantitative Trait Loci, and we will see later on what these are that drive the trait variation. And the second parts, well, that has more to do with optimization theory, it's combining the desired QTL alleles in efficient ways. Okay, so as said, we will mainly talk about genotype to phenotype models, as this is, I think, the first and most general step that we need to go through before we can improve anything at all in plant breeding. And we will restrict ourselves, let's say, to the kind of working horse in plant breeding, that is the mixed model. The mixed model is, say, like a linear model that has both fixed and random terms. And we will have more to tell you about that later. And related to this use of mixed models, we will also have, let's say, a little bit of a view on estimation procedures for what we call the genetic values. And the genetic values are, well, those things that allow us to select between different plants and find the best ones. And the best ones for crossing and transmitting, let's say, desirable alleles and genes to the next generations. And we also have a view not only on the breeding values which characterize individuals, but we also look at parameters like genetic variances, heritabilities et cetera that are more specific for a totality or a population of individual plants.
4:05
We see that an important and major activity in plant breeding is selection. And over, that's still about 25 years ago, phenotypic selection was the major way in which we could find, identify our superior plants, our superior parents. And what we hoped is that when we would identify superior parents and we crossed those, that we would get superior offspring. And of course, the success of getting superior offspring from a cross between superior parents depends on how well a phenotype is informative of the underlying genetic factors that, let's say, create, cause the superiority. Nowadays, we have more or less replaced that paradigm of phenotypic selection by what we call marker assisted or genome enabled selection. And in both cases, the ideas that we identify superior parents on the basis of genes and QTLs. And these genes and QTLs, well, they form the basis of our breeding values that what distinguish a good genotype for a bad genotype, a good plant from a bad plant. So the superior parents are now crossed based on marker profiles or DNA profiles. And the best way to guarantee success using those markers is if we are able to identify the molecular markers that are close or inside the casual genes. Although, of course, that is not so easy. Both, our classical phenotype selection and our current, let's say, more modern marker assisted and genome enabled selection procedures require the use of appropriate genotype to phenotype models. And once again, we will mainly talk about those in terms of mixed models.
Hide

Introductory statistical genetics for plant breeding 1

Embed in course/own notes