In Increasing accuracy of genomic predictions: fr from SNP chips - - PowerPoint PPT Presentation

▶
in increasing accuracy of genomic
SMART_READER_LITE
LIVE PREVIEW

In Increasing accuracy of genomic predictions: fr from SNP chips - - PowerPoint PPT Presentation

May 03, 2023 39 likes •387 views

In Increasing accuracy of genomic predictions: fr from SNP chips to sequence data Daniela Lourenco Shogo Tsuruta and Ignacy Misztal 6/12/2020 52nd Beef Improvement Federation Research Symposium and Convention Genomics in in animal and pla

slide-1
SLIDE 1

In Increasing accuracy of genomic predictions: fr from SNP chips to sequence data

Daniela Lourenco

Shogo Tsuruta and Ignacy Misztal

6/12/2020 52nd Beef Improvement Federation Research Symposium and Convention

slide-2
SLIDE 2

Genomics in in animal and pla lant breeding

2

Use of DNA markers

  • Construct genetic relationships
  • Parentage determination
  • Identification of genes/QTL
slide-3
SLIDE 3

SNP - Sin ingle nucleotide poly lymorphisms

http://www.thinnergene.com/about-thinnergene/genetics-101/

Individual 1 Individual 2

3

“SNPs have become the bread-and-butter of DNA sequence variation” (Stonecking, 2001)

slide-4
SLIDE 4

What are SNPs doin ing?

4

gene gene gene

  • Identify genes
  • Track relationships
  • Parentage determination
slide-5
SLIDE 5

How many SNP in in the genomic test?

5

https://www.sciencelearn.org.nz /images/2543-dna-precipitate https://www.lihs.cuhk.edu.hk/en-us/corefacilitiesandresources/corefacilities/geneexpressionprofiling,analysisandgenotyping.aspx
  • Livestock, poultry, aquaculture: 3,000 - 800,000 SNP
slide-6
SLIDE 6

Genomic Selection

GEPD

Genomic Pedigree Performance

Progeny data

slide-7
SLIDE 7

Adoption of genomic testing

780,000

http://www.angus.org/AGI/default.aspx

3,400,000

http://www.holsteinusa.net/programs_services/backgrounds.html https://www.usjersey.com/AJCA-NAJ- JMS/AJCA/AnimalIdentificationServices/HerdRegister.aspx

460,000

http://sesenfarm.com/raising-pigs/

~50,000/line 2,000 - 10,000

www.sheepcrc.org.au/

50,000 - 70,000

7

~50,000/line

https://www.cobb-vantress.com/
slide-8
SLIDE 8

Benefits of genomic testing in in beef cattle

0.29 0.34 0.23 0.39 0.38 0.29 0.47 0.40 0.35 BW WW PWG

Angus

BLUP ssGBLUP14 ssGBLUP17

Average gain in accuracy 52k animals = 25% 335k animals = 36% Genotyped for 50k SNP

Lourenco et al., 2018

8

The gain in accuracy is because genomics is a non-redundant piece of information!

slide-9
SLIDE 9

Benefits of genomic testing

9 https://www.angus.org/AGI/GenomicEnhancedEPDs.pdf

Cole, 2020

slide-10
SLIDE 10

Benefits of genomic testing

Genetic gain = (selection intensity * accuracy * genetic SD) / generation interval

10

slide-11
SLIDE 11

Benefits of genomic testing in in dairy ry cattle

  • Increase in genetic gain
  • ~50 to 100% for yield traits
  • 3- to 4-fold for traits with low heritability (fertility)

11

Genotyped for ~ 60k SNP

slide-12
SLIDE 12

Benefits of genomic testing in in pig igs

12

Selection Index

2006 2013/2014 2019

ssGBLUP

Genotyped for ~ 60k SNP

slide-13
SLIDE 13

Benefits of genomic testing

  • Increase in accuracy
  • Number of genotyped animals
  • Amount of performance records
  • Heritability of the trait
  • How informative the genomic test is
  • 50k SNP in beef cattle

Is the genomic test informative enough? Do we have enough SNPs in the test? Are we looking at the right SNPs?

13

slide-14
SLIDE 14

About the rig ight SNP

  • Is the genomic testing informative enough?
  • Do we have enough SNPs in the test?
  • Are we looking at the right SNPs?

54k SNP cattle 38k SNP 54k SNP pigs 44k SNP 54k SNP chicken 39k SNP

14

slide-15
SLIDE 15

Sequence in information

http://mtc.science/playing-around-with-ngs-step-by-step

50k SNP 30M SNP

15

slide-16
SLIDE 16

SNPs capture rela lationships at genes

  • Expected relationship between a

bull and its grandparents?

  • Observed relationships based on

proportion of alleles shared between a bull and its paternal grandsire and maternal grandsire?

Jared Decker, University of Missouri

25.8% 15.4%

With sequence data, we may find SNPs that give more precise information about genes because they may be closer to genes

16

slide-17
SLIDE 17

Sequence in information

~ 50k SNP ~ 30M SNP

  • Trying to get a more precise information about the genes
  • More SNPs to cover larger areas of the DNA
  • Select SNP that are closer to genes for traits of interest

17

slide-18
SLIDE 18

Sequencing animals

  • Becoming cheap
  • Sequence a small portion of animals
  • Imputation from 50k to sequence (filling the gaps)
  • 18,000 to 107,000 pigs
  • Sequenced 2% of the animals
  • Imputation to 20M to 30M
  • Imputation accuracy of 94% - 98%

18

slide-19
SLIDE 19

Largest sequence data in in pig igs

  • Gain in accuracy
  • Multibreed
  • Persistency of accuracy
  • Up to 100,000 sequenced/imputed pigs
  • 20M to 30M SNP

Do we work with all ~30M SNP?

19

slide-20
SLIDE 20

Min ining sequence data

Adapted from: https://www.broadinstitute.org/visuals/explainer-genome-wide-association-studies

Traits

20

slide-21
SLIDE 21

Aguilar et al., 2019

Birth weight in Angus

Min ining sequence data

  • Causative SNP
  • SNP give information

about birth weight

21

slide-22
SLIDE 22

Selecting causative SNP

  • Issues
  • Effective population size
  • Number of animals with sequence and phenotypes
  • Large data – more confidence in finding causative SNP
  • Small data – more difficult to identify

– more susceptible to errors

22

slide-23
SLIDE 23

Selecting causative SNP

10 genes

  • f equal

effect 500 SNP to trace 10 genes

23

Small population

Pocrnic et al., 2020

slide-24
SLIDE 24

Selecting causative SNP

Ne = 60 Animals = 18000 Ne = 60 Animals = 6000 Ne = 600 Animals = 6000 Pocrnic et al., 2020

24

slide-25
SLIDE 25
  • Small data: not enough power to discover which SNP are truly causative
  • Methods are also important

Selecting causative SNP

Classical GWAS EMMAX ssGWAS Mancin et al., 2020

25

slide-26
SLIDE 26

Usin ing selected SNP to compute GEPD

  • Hanwoo cattle from South Korea
  • 545k Marbling score records
  • 1.3M animals in pedigree
  • 1160 genotyped animals
  • Imputation: 50k to 777k to 11.1M SNP
  • 321k SNP selected out of 11.1M sequence SNP

en.prnasia.com

Jang et al., 2020

26

slide-27
SLIDE 27

Gain in in accuracy with selected SNP

0.19 0.19 0.27 0.16 0.16 0.26 0.18 0.18 0.27 BayesR GBLUP ssGBLUP 50k SEL 50k+SEL

Jang et al., 2020

No gain in accuracy with sequence SNP Small data: animals with SNP and records GBLUP/BayesR = ~ 1200 records ssGBLUP = 545k records

27

slide-28
SLIDE 28
  • VanRaden et al. (2017)
  • 27k US Holstein bulls
  • Selected 16.6k SNP based on effect size on 33 traits
  • Added to the 60k SNP chip
  • 0.05
  • 0.03
  • 0.01

0.01 0.03 0.05

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

Gain in accuracy by using selected sequence SNP Average

Gain in in accuracy with selected SNP

Stature Milk Fat Protein

28

slide-29
SLIDE 29
  • Data from VanRaden et al. (2017) – Bayes A
  • 27k US Holstein bulls
  • Selected 16.6k SNP based on effect size on 33 traits
  • Added to the 60k SNP chip

Gain in in accuracy with selected SNP

29

  • Fragomeni et al. (2019) – ssGBLUP

0.828 0.869 0.854 0.871 BayesA ssGBLUP

Stature in US Holsteins

60k 60k+SEL

slide-30
SLIDE 30

Sim imulated data – real genes

  • Simulated population
  • 180,000 animals
  • 30,000 genotyped
  • 60,000 SNP
  • 100 genes (QTN)
  • Single-step genomic predictions based on
  • SNP
  • SNP + QTN
  • QTN

30

slide-31
SLIDE 31

Accuracy wit ith sim imulated genes

Fragomeni et al. (2017)

31

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 60k SNP - single-step 60k SNP + 100 QTN - single-step 60k SNP + 100 QTN - single-step "true" weights 100 QTN - single-step

0.49 0.53 0.89 0.99 0.52 0.75 100% 10%

slide-32
SLIDE 32

Accuracy wit ith sim imulated genes

32

Fragomeni et al. (2017)

60k + 100 QTN – ssGBLUP APY – True weights

QTN

  • Greatest accuracy if
  • all genes
  • locations are known
  • weights are optimal
slide-33
SLIDE 33

Is Is it it possible to in increase accuracy of genomic EPD wit ith sequence data in in real cattle populations?

SNP closely linked to genes SNP that turn genes on and off We don’t know all the genes Knowing a bit may help a bit Amount of information they give Lots of work in progress

33

slide-34
SLIDE 34

Acknowledgements

34