Shulou(Shulou.com)06/01 Report--

Today, I would like to talk to you about the answers to FAQ, a frequently asked question about how to make genome selection. Many people may not know much about it. In order to make you understand better, the editor has summarized the following for you. I hope you can get something from this article.

Frequently asked questions on Genome selection FAQ

The following describes the common problems of GS, translation and learning.

1. I heard that if there are 1000 genotypic and phenotypic values of animals, genome-wide selection can be made and the accuracy can be maintained for many generations. is this true?

Fake! As the selection generation increases, the accuracy decreases, so the sequencing population needs to be constantly updated. The reference group needs to be constantly updated, and phenotypic individuals after sequencing can be added.

2. How many individuals do I need to sequence to achieve a significant improvement in the accuracy of the evaluation?

In cows, each bull has many offspring, each group is larger, and the sequenced individuals are about 2000. For species with relatively few offspring and low heritability, it is necessary to increase the number of populations. Generally speaking, the number of reference groups must reach at least 600 in order to achieve a significant improvement in accuracy.

3. Which is better, rrBLUP or GBLUP, two-step method (first estimating the SNP effect value, then calculating the breeding value) or one-step method (directly estimating the breeding value)?

RrBLUP and GBLUP are equivalent. It is easier to use rrBLUP to estimate variance components and to calculate GBLUP. When the weights is known, the two are equivalent.

4, is the weight of SNP important?

Initial studies have shown that the weight of SNP is very important, but with the increase of SNP density, setting the weight of each SNP to the same is suitable for most traits. (that is, it is assumed that all SNP have the same variance distribution, the effect values are known, and they are all controlled by minor polygenes). For traits controlled by only a few major genes (QTL), the effect may not be very good.

5. Can one group be used as a reference group to predict the GEBV of another population?

This depends on the genetic relationship between the two groups. If many parents or ancestors of one group are parents or ancestors of another group, the accuracy of the evaluation is higher, and if there are fewer parents or ancestors, the accuracy of the evaluation is lower. In extreme cases, there is no relationship between the two groups, and the accuracy of the evaluation is even lower than the traditional EBV (pedigree-blup).

6. Can the value of SNP effect obtained from the reference group of one group be used to evaluate other groups?

No, the effect value of SNP can only be used in similar populations, and we get more accurate additive variance components through SNP. The effect value of SNP can not be applied to other unrelated populations.

7. How much of the additive effects can be explained by genes or linked SNP?

In general, the traits controlled by multiple genes are generally 5% to 20%, but the latest research shows that the proportion can reach 50%. However, these studies can find some G relation matrices in some individuals with less correlation.

8, what if the candidate group comes from many groups (pedigrees)?

Then this candidate group can predict many other pedigrees (because the consanguinity of the candidate group is complex) and the simple hybridization between them. But the G matrix needs to be corrected and standardized to prevent the deviation of the estimation. When estimating families, the accuracy will be relatively reduced.

9. If we sequence multiple generations, will it help to improve accuracy?

It may or may not. With the selection, the accuracy of the G matrix will decrease, because the background of the previous candidate group has changed due to the selection.

10, what's the difference between G matrix and A matrix?

If the pedigree is more complete, the algebra is more, and the structure is single, then the A matrix and G matrix are basically the same, and the standard deviation of the difference can be less than 0.04.

There are many ways to build a G matrix. Is there a big difference between them?

For species with a large number of offspring (such as dairy cows), almost all G-matrix construction results get the same GEBV. When the population is small, or the pedigree and genome are merged (one-step method), if the two (An and G) are not corrected and standardized, the GEBV may be biased and the accuracy will be reduced.

12. Traditional animal models have many functions, such as correction, maternal effect, group grouping (unknown parent groups), hybrid group and so on. Can genome selection achieve these functions?

All of this can be achieved using genome selection. Imagine that GBLUP only replaces A matrix with G matrix relative to ABLUP. It can perform repetitive force analysis, maternal effect and other functions that traditional animal models can do. In the part of group grouping, there are few studies on GBLUP.

13. Will the variance component estimated by GBLUP be higher than that estimated by ABLUP?

If the pedigree is correct and the G matrix is corrected, then they should be similar. The error of variance components estimated by G matrix may be smaller. If genomic information is available, genomic information can be used to estimate variance components directly without using pedigree.

14. Will the accuracy of estimation be improved if high-density SNP chips are used?

It is not clear that the accuracy of estimation using a 50K chip is quite good, so there is no limit to the effect of continuing to increase SNP density. If the group is relatively small, and the chip density is relatively high, it will cause a false increase.

15. If we have complete genomic data, identify all susceptible SNP loci (associated SNP), and estimate the effect values of these SNP in a population or a mixed population, can we accurately predict the breeding value of any individual?

This is a good problem, there is a complex network relationship between genes, and the relationship between susceptible genes and traits is usually non-linear, in other words, you have some susceptible SNP, you want to get phenotypic value, but also consider the environment, interaction and so on. Of course, it also has certain reference significance for traits with high heritability or controlled by major effect genes (QTL).

16, if this is the result of the previous question, then why does the animal model work so well?

In the short-term assessment, the effect is better, and the animal model evaluation is often aimed at the next generation, and the environmental change table is small.

There is a lot of information about the genome. if we get enough data, can we make a breakthrough?

Taking into account all the parity, the number of offspring of a pair of parents is generally no more than 100, and these offspring will have mutations.

18, animal breeding, what will be the next technology of genome selection?

No one knows, at present, genome selection has greatly improved the accuracy, but there is room for further use, such as hybridization or gene-environment interaction.

19. What is the most important thing in genome selection?

Pay attention to the accuracy of phenotypic and genotypic data. Studies have shown that if the quality of genotypic data is poor, it will greatly affect the analysis results.

After reading the above, do you have any further understanding of the answers to FAQ, a frequently asked question on how to make genome selection? If you want to know more knowledge or related content, please follow the industry information channel, thank you for your support.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.