Genome Wide Association Study of two phenology traits in apple

Hélène Muranty, INRA-Angers, France

Acknowledgements Conference organizers

Co-authors• C. Denancé, D. Leforestier, E. Ravon, A. Guyader, R. Guisnel, L. Feugey, F.

Laurens, C.-E. Durel• J. Urrestarazu, S. Tartarini, L. Dondini, R. Gregori• M. Lateur, P. Houben• J. Sedlak, F. Paprstein • M. Ordidge • H. Nybom, L. Garkava-Gustavsson• M. Troggio, L. Bianco, R. Velasco• M.C.A.M. Bink, W. Kruijer

INRA-Clermont-Fd, Gentyane platform: C. Poncet, …

Why GWAS ?

• Discover and quantify effects of genomic regions associated to complex traitssubstantially increase resolution by using collections

of unrelated individuals

candidate genesmarkers for selectionorigin of favorable alleles Yu and Buckler (2006)

many generations

Individual M1 M2 M3 M4 Phenotype

1 2 2 2 1 202 1 0 0 0 83 2 2 0 0 244 1 1 2 1 45 1 0 0 0 146 1 1 2 1 12

-log p

Chromosomal location

GWAS components

Scoring Detection

Mackay et al (2009)

Very high density genotypingAffymetrix Axiom_Apple480k array

Why phenology traits ?

• target cultivar development / growing season length in production areas

• develop cultivars able to face climate change challenges

Data available - MaterialCountry SizeF – INRA (CC) 278B – CRA-W 229UK – U. Reading (CC) 294CZ - RBIPH 178I – UNIBO (CC) 179SW – SLU 162

1168

6 collectionsold and local dessert apple cvrs

Phenotypes

trait scale note

flowering period

1 (Extremely early) -> 9 (Extremely late)

comparison to reference cultivars

picking date days from 1st January

picking period 1 (Extremely early) -> 9 (Extremely late)

comparison to reference cultivars

Phenotypic analysis

site Flowering period Picking Date Picking period

# data/gt h² # data/gt h² # data/gt h²

F - INRA 3.0 0.88 3.9 0.96 1.8 0.86

B - CRA-W 4.9 0.88 1 - 4.3 0.87

UK - U. Reading ? - 2.0 0.88 1? -

CZ - RPIPH 5.0 0.85 5.1 0.92

I - UNIBO 7.6 0.84 2.1 0.96 6.6 0.94

SW- SLU 3.0 0.81 2.9 0.98 2.9 0.98

all sites 0.82 0.94 0.89

Heritability of the means

genotypic means adjusted for • year effects: collection per collection analysis• (site x year) effects: all collections together

= =

NMHom: No Minor Homozygous

2.4%

Genotypes: quality control and filtering

275K

Additional filtering pipeline

technical replicates (GoldenDel)biological replicatesMendelian consistency

mapping progeniesparent-offspring pairs

NMHom: No Minor Homozygous 12K

correct Poly High Resolution 360KUnexpectedHeterozygosity 11K

criteria from SNP Polisher visual scoring ~1600 SNP good/poorlogistic regression -> quality prediction

Affymetrix Axiom_Apple480k array

487K

SNP Polisher

Samples: DQC > 0.82 and CallRate > 97%

Unexpected Heterozygosity

2.8%

correct PHR73.8%

Physical map: present drawbacks• Scaffold orientation on LG undetermined• arbitrary 1000bp between scaffolds on LG• some scaffolds attributed to LG without position•many SNP (~25%) on scaffolds not attributed to any

LG LG0 (LG18)

• some SNP from previous arrays not located on the present physical map LG20

GWAS: Model choice to avoid false positives

Y = µ + SNP + e Y = µ + Q + SNP + e Y = µ + K + SNP + e Y = µ + Q + K + SNP + e

Population structure in apple

(3.9 %)

(2.9

%)

Flowering periodCRA-W INRA NFC

RBIPH SLU UNIBO

SNP + Q + K model

Flowering period: advanced model

INRA NFC RBIPH

SNP + Q + K model + SNP cofactorsExtended BIC model selection criteriaMLMM Ségura et al (2012)

LG9

LG9 x 2LG12

LG11LG9

Flowering period: all collections

SNP + Q + K model

SNP + Q + K model + SNP cofactorsExt BIC best modelMLMM Ségura et al (2012)

Flowering period:Affx-113839215 : LG9 @ 1293620

Picking dateINRA NFC

SLU UNIBO

SNP + Q + K model

Picking date: INRA+NFC+SLU+UNIBO

Comparison to previously detected regionstrait Chr regions in GWAS

(Mb)region in QTL analysis

(cM (Mb))comment reference

Flowering period

9 1.3 -1.6 0.4 (0.6) Belrène, 2 years Celton et al (2011)2.5

Allard et al Eucarpia Fruit 201516.9

Picking date 3 28.6 – 30.3 53.2 (26.0) Braeburn, 3 site-year comb

Chagné et al (2014)

44.3 75 Discovery Liebhard et al (2003)

QTL position IC length ~10 cM

74kb

676kb

18kb222kb

127kb

Conclusions & Perspectives

• GWAS can detect already known QTL = proof of concept• Variation explained by kinship (+ structure) =

small effect QTLs undetectable genomic prediction• Look for candidate genes• markers for selection• origin of favorable alleles

Welcome in Angers, June 22-24 2016Rosaceae Genomics Conference 8

Thank you for listening