Genome-wide association study of a diverse grapevine panel to uncover the genetic architecture of numerous traits of interest

HAL Id: hal-02734488

https://hal.inrae.fr/hal-02734488

Submitted on 2 Jun 2020

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Genome-wide association study of a diverse grapevine

panel to uncover the genetic architecture of numerous

traits of interest

Timothée Flutre, Roberto Bacilieri, Gilles Berger, Yves Bertrand, Jean-Michel

Boursiquot, Agota Fodor, Thierry Lacombe, Valerie Laucou, Amandine

Launay, Loïc Le Cunff, et al.

To cite this version:

Timothée Flutre, Roberto Bacilieri, Gilles Berger, Yves Bertrand, Jean-Michel Boursiquot, et al.. Genome-wide association study of a diverse grapevine panel to uncover the genetic architecture of numerous traits of interest. 12. International Conference on Grapevine Breeding and Genetics, Jul 2018, Bordeaux, France. �hal-02734488�

Grapevine Breeding and Genetics  17/07/2018

Genome-wide association study

of a diverse grapevine panel

to uncover the genetic architecture

of numerous traits of interest

J-M Boursiquot, A. Fodor, T. Lacombe, V. Laucou, A. Launay, L. Le Cun, C. Romieu, P. This, J-P. Péros, A. Doligez

Multiple changes and challenges

Reduce pesticides Adapt to climate change

ref. T◦_{in 1970 pred. T}◦_{in 2055 pred. T}◦_{in 2085}

ARPEGE model

Major questions to biologists:

1. how to phenotype the eco-physiological processes of interest? 2. what are their genetic architectures?

Multiple changes and challenges

Reduce pesticides Adapt to climate change

ref. T◦_{in 1970 pred. T}◦_{in 2055 pred. T}◦_{in 2085}

ARPEGE model

Major questions to biologists:

1. how to phenotype the eco-physiological processes of interest? 2. what are their genetic architectures?

Diversity panel of Vitis vinifera L. from Domaine de Vassal

Field layout at Domaine du Chapitre

2009: overgrafton Marselan (control)

I 5 complete

randomized blocks

I each genotype has 1

replicate per block

c

Intense phenotyping eort

I Traits: mean berry weight; mean bunch weight, length and

compactness; pruning weight and number of woody shoots; malate, tartrate, shikimate; δ13_C

I Additional covariates: vigour, sanitary status I No irrigation

2014-2015

I Traits: mean berry weight; δ13C; β-damascenone and pDMS;

polyphenols (Pinasseau et al., 2017)

I Treatment: with or without irrigation

Intense phenotyping eort

I Traits: mean berry weight; mean bunch weight, length and

compactness; pruning weight and number of woody shoots; malate, tartrate, shikimate; δ13_C

I Additional covariates: vigour, sanitary status I No irrigation

2014-2015

I Traits: mean berry weight; δ13C; β-damascenone and pDMS;

polyphenols (Pinasseau et al., 2017)

I Treatment: with or without irrigation

Mean berry weight: exploratory analysis of phenotypes

Mean berry weight: exploratory analysis of phenotypes

Mean berry weight: exploratory analysis of phenotypes

Dual genotyping

I GrapeReSeqmicroarray(Illumina): 12k SNPs after QC I GBSwith ApeKI enzyme (Keygene): 120k SNPs after QC I Combined: 90k SNPs with LD < 0.9 and MAF > 0.01

11k SNPs 90k SNPs

Dual genotyping

I GrapeReSeqmicroarray(Illumina): 12k SNPs after QC I GBSwith ApeKI enzyme (Keygene): 120k SNPs after QC I Combined: 90k SNPs with LD < 0.9 and MAF > 0.01

Statistical analysis of phenotypic data

y = X β + Zg +  with g ∼ N (0, σ2gId);  ∼ N (0, σ2Id)

I y: phenotypic observations

I β: eects of known factors, modeled as "xed"

I g: total genotypic values, modeled as "random" I : errors

I H2 = σ2g

σg2+(σ2/#rep): broad-sense heritability(of means)

y = X β + Za+ 0 witha∼ N (0, σ_a2A);  ∼ N (0, σ02Id)

I A: kinshipmatrix of additive genetic relationships I a: additive genotypic values(a.k.a. breeding values) I h2 = σ2a

Statistical analysis of phenotypic data

y = X β + Zg + with g ∼ N (0, σ2_gId);  ∼ N (0, σ2Id)

I y: phenotypic observations

I β: eects of known factors, modeled as "xed"

I g: total genotypic values, modeled as "random" I : errors

I H2 = σ2g

σg2+(σ2/#rep): broad-sense heritability(of means)

y = X β + Za+ 0 witha∼ N (0, σ_a2A);  ∼ N (0, σ02Id)

I A: kinshipmatrix of additive genetic relationships I a: additive genotypic values(a.k.a. breeding values) I h2 = σ2a

Estimation of heritabilities

H2: higher, better →g well approximated by its BLUP h2: higher, better → σ2_a large enough for selection purposes

Statistical analysis of genotypic values

BLUP(g) = 1µ + mpβp + u + 

I βp: eect of the pth SNP → test if βp=0

I u: polygenic eect with kinship matrixK ∝ MMT

Multi-SNP: explicit modelling of the genetic architecture BLUP(g) = 1µ + Mβ+ 

I fully polygenic: all β_p6=0

I major QTLs only: few β_p6=0 and all others = 0

Statistical analysis of genotypic values

BLUP(g) = 1µ + mpβp + u + 

I βp: eect of the pth SNP → test if βp=0

I u: polygenic eect with kinship matrixK ∝ MMT

Multi-SNP: explicit modelling of the genetic architecture BLUP(g) = 1µ + Mβ+ 

I fully polygenic: all β_p6=0

I major QTLs only: few β_p6=0 and all others = 0

Statistical analysis of genotypic values

BLUP(g) = 1µ + mpβp + u + 

I βp: eect of the pth SNP → test if βp=0

I u: polygenic eect with kinship matrixK ∝ MMT

Multi-SNP: explicit modelling of the genetic architecture BLUP(g) = 1µ + Mβ+ 

I fully polygenic: all β_p6=0

I major QTLs only: few β_p6=0 and all others = 0

Estimation of hybrid genetic architectures

PVE: proportion of variance of total genotypic values explained by thepolygenic component and themajor QTLeects

I higher → better to predict genotyping values

PGE: proportion of PVE explained only bymajor QTL eects

I higher → better to identify candidate genes

trait #SNPs med(#QTLs)

mbw 11k 31 [0,169]

mbw 90k 14 [2,115]

I importance of

genotyping densication

I large amount of genetic

variance from polygenic component

Estimation of hybrid genetic architectures

PVE: proportion of variance of total genotypic values explained by thepolygenic component and themajor QTLeects

I higher → better to predict genotyping values

PGE: proportion of PVE explained only bymajor QTL eects

I higher → better to identify candidate genes

trait #SNPs med(#QTLs)

mbw 11k 31 [0,169]

mbw 90k 14 [2,115]

I importance of

genotyping densication

I large amount of genetic

variance from polygenic component

Mean berry weight: SNP-by-SNP versus multi-SNP

Mean berry weight: SNP-by-SNP versus multi-SNP

Mean berry weight: SNP-by-SNP versus multi-SNP

Mean berry weight: SNP-by-SNP versus multi-SNP

I PVE = 0.668 [ 0.613, 0.735 ]d

Mean berry weight: selected SNPs

Mean berry weight: selected SNPs

SNP #2 at ≈ 29.9 Mb on chr14

\

Pr(βp6=0) = 0.999 ; dPVEp=0.074 ; bβp= −0.159 ; CI95%= [−0.202, −0.117] location: promoter of Vitvi14g02008, uncharacterized

Prospects with the panel

I improved phenotyping of berry physiology(poster 49);

tolerance topathogens (poster 57)

I phenotyping inmultiple sites andgreenhousesto study GxE

Genotyping:

I capture-based sequencing of GBS-dened SNPs I search for traces of selection

Take-home message

With

dense

genotyping and

multi-SNP

models,

the

diversity panel

of V. vinifera L.

from INRA Montpellier

allows estimating the

genetic architecture

of

numerous traits of interest,

to help design ecient

breeding

strategies.

I data and reproducible analyzes: available upon publication I contact: Agnès Doligez (agnes.doligez@inra.fr)

Acknowledgments

I DAAV team from the UMR AGAP

I Vassal-Montpellier grapevine biological resources center I SouthGreen bioinformatics platform (notably B. Pitollat) I AGAP genotyping platform (notably P. Mournet)

I GenoToul sequencing center