Detecting directional and epistatic selection from candidate genes: methodological improvements and a case study of European beech

HAL Id: hal-01204227

https://hal.archives-ouvertes.fr/hal-01204227

Submitted on 5 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.

Detecting directional and epistatic selection from candidate genes: methodological improvements and a

case study of European beech

Katalin Csillery, Giovanni Giuseppe Vendramin, Santiago C González-Martínez, Bruno Fady, Sylvie Muratorio

To cite this version:

Katalin Csillery, Giovanni Giuseppe Vendramin, Santiago C González-Martínez, Bruno Fady, Sylvie Muratorio. Detecting directional and epistatic selection from candidate genes: methodological im-provements and a case study of European beech. SMBE Satellite meeting SMBEBA 2015 ”Investi-gating biological adaptation with NGS: data and models”, May 2015, Hameau de l’étoile, France. 1 p. �hal-01204227�

Detecting directional and epistatic selection from candidate genes:

methodological improvements and a case study of European beech

K C

SILLÉRY

1

, H L

ALAGÜE

1,2

, GG V

ENDRAMIN

2

, SC G

ONZÁLEZ

-M

ARTÍNEZ

3

, B F

ADY

1

, S O

DDOU

-M

URATORIO

1

INRA Avignon (FR)

, Inst. of Biosciences & BioResources (IT)

, INIA Madrid (SP)

, with financial support from: ERANET-BiodivERsA: LINKTREE & TIPTREE

S

UMMARY

S

-

•

accounting for the uncertainty of haplotype inference in F

outlier tests

•

re-discoveing Ohta’s test of epistatic selection for candidate genes

A

Fagus sylvatica

• sampling at a short spatial scale with sharp environmental differences

• SNPs from candidate genes potentially involved in climate response

See more details in: K Csilléry, H Lalagüe, GG Vendramin, SC González-Martínez, B Fady and S Oddou-Muratorio 2014 Detecting local adaptation and epistatic selection in climate related candidate genes at a short spatial scale in European beech

(Fagus sylvatica L.) populations. Molecular Ecology 23: 4696-4708

R

ESULTS

: D

IRECTIONAL SELECTION

●●● ● ● ● ●●●●●● ● ● ● ● ● ● ● ● ●● ● ●●●●●●●●● ● ● ● ● ●●● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ●●● ●● ● ●● ●●● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ●_● ●● ● ● ●●● ●●●●●●●●● ● ● ●_●●●●●●●●●● ● ●●● ● ● ● ● ●● ● ● ● ● ● ●●●● ● ● ● ● ● ● ● ● ● ● ● ●_● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ●●●●● ● ●●● ● ● ● ● ● ● ●●●●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●●●● ● ● ● ● ● ● ● ● ●●●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ●● ● ● ●● ● ● ● ● ●● ● ● ● ●● ● ●● ● ● ● ● ●● ● ● ● ● ● ●●●_●●●● ●●●●●● ● ● 1.0 1.5 2.0 2.5 3.0 3.5 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 SNP level Bayes Factor Fs t 142.328 23_1.787 ● North South A 0.5 1.0 1.5 2.0 2.5 3.0 0.00 0.05 0.10 0.15 Gene level

Median Bayes Factor

Median F s t ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 134_2_2 ● North South B 0.5 1.0 2.0 5.0 10.0 0.004 0.008 0.012 0.016 Gene 23_1 Bayes Factor Fs t North C 1 2 5 10 0.02 0.04 0.06 0.08 0.10 Gene 142 Bayes Factor Fs t North D 1 2 5 10 20 0.004 0.008 0.012 0.016 Gene 134_2_2 Bayes Factor Fs t North E ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●_● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 1 2 5 10 20 0.05 0.15 0.25 Gene 58 Bayes Factor Fs t ● South F

C

:

• outlier detection at the SNP (A) vs

gene level (B) revealed different loci

under selection

• gene level outlier detection was

strongly influenced by uncertainty in

haplotype reconstruction

(C-F)

• uncertainty of haplotype inference

can be accounted for by averaging

Bayes factors over many possible

phase reconstructions

(B)

K

EY REFERENCES

1

Foll and Gaggiotti. Genetics 180.2 (2008): 977-993; 2Stephens and Scheet. AJHG 76 (2005):449-462; 3Ohta. PNAS 79.6 (1982): 1940-1944; 4Mackay. Nat. Rev.

Genet. (2013); 5Hansen. Evolution 67.12 (2013): 3501-3511; 6Lander et al. Mol. Ecol. 20.24 (2011): 5182-5196; 7Lehner. Trends in Genet. 27.8 (2011): 323-331.

F

IRST AUTHOR

’

S CONTACT

Present address:

INRA Avignon, UR629, France

Address from Oct 2015:

ETH Zürich, ACE, Switzerland

Web https://sites.google.com/site/katalincsillery/

Email [email protected]

R

ESULTS

: E

PISTATIC SELECTION

North vs South 10 100 110_1 110_3 125 129 130 131 133 134_2_2 14 142 145_2 148_1 150_2 154_1 154_2 155_2 155_3 156 17 19 20 21 23_1 24 27 30_2 33 39 4 47_1 50 51_2 52_1 52_2 58 60 61_2 62_1 66 68 7 70 73 80 88_1 88_2_1 88_2_2 91_2 92 98_1 99 10 100 110_1 110_3125 129 130 131 133 134_2_214 142 145_2 148_1 150_2 154_1 154_2 155_2 155_3156 17 19 20 21 23_124 27 30_233 394 47_150 51_2 52_1 52_258 60 61_2 62_166 687 70 73 80 88_1 88_2_1 88_2_291_2 92 98_199 Nor th South A 0 2 5 8 10 14 16 28 32 64 High vs Low 10 100 110_1 110_3 125 129 130 131 133 134_2_2 14 142 145_2 148_1 150_2 154_1 154_2 155_2 155_3 156 17 19 20 21 23_1 24 27 30_2 33 39 4 47_1 50 51_2 52_1 52_2 58 60 61_2 62_1 66 68 7 70 73 80 88_1 88_2_1 88_2_2 91_2 92 98_1 99 10 100 110_1 110_3125 129 130 131 133 134_2_214 142 145_2 148_1 150_2 154_1 154_2 155_2 155_3156 17 19 20 21 23_124 27 30_233 394 47_150 51_2 52_1 52_258 60 61_2 62_166 687 70 73 80 88_1 88_2_1 88_2_291_2 92 98_199 High Low B 0 2 4 8 15 26 28 32 68 North 39 142 52_1 52_2 98_1 61_2 68 91_2 South 10 145_2 155_2 148_1 150_2 23_1 68 39 98_1 52_1 52_2 91_2 142 61_2 80 155_3 High 39 142 52_1 52_2 61_2 91_2 98_1 Low 142 148_1 39 52_2 52_1 80 91_2 98_1

Gene pairs “light-up” in red if they contain at least an SNP that show a signal of epistatic selection between them. Genes in the diagonal show within-gene epistatic selection signal. Networks are drawn between genes that show a unique between-gene epistatic selection signal.

W

?

Key genes with unique between gene epistatic selection signal in North/High (N/H) and South/Low (S/L) population pairs:

• (N/H) Gene 68 is connected 61_2 and 142 via two non-synonymous SNPs. Gene 61_2 is a member of the heat shock protein 70 family and 68 catalyzes glycolysis, both play a key role in stress response.

• (S/L) Gene 50’s SNP was situated in a 3’UTR region and the gene codes a major transcription factor in response to abiotic stress and has been shown to respond to cold temperatures.

• (S/L) Gene 80 regulates stomatal closure (key importance in re-sponse to drought) and has been suggested to play a role in dor-mancy.

• (S/L) Genes 148_1 and 145_2 are well-known budburst candi-date genes.

W

?

Ohta’s test has been relatively little used and most studies found no signal of epistatic selection, so why did it work here?

• Recent selection: F. sylvatica populations re-colonized Mont Ventoux about five generations ago6 and since have been ex-posed to sharp environmental differences

• Functionally related genes favored the build-up and mainte-nance of LD due to epistatic selection

• Samping from sharply different environments: 0.23% of the SNP pairs showed evidence of epistatic selection, with nearly 80% of them being within genes. However, most epistatic in-teractions unique to a population pair (N, S, H, or L) were ob-served between different genes. Indeed, most systematically mapped epistatic interactions between different genes bring

new functionality that may only be advantageous in a

partic-ular environment7

M

ATERIALS

& M

ETHODS

S

0 1 2 3 4 5 km

NL: North Low, NH: North High, SL: South Low, SH: South High

D

:

F

outlier test at the SNP and candidate gene levels with Bayescan

For gene level tests, haplotype phase was estimated using PHASE

D

:

Following Ohta

, we decomposed the variance of linkage

disequilib-rium within a subdivided population into between and within

popu-lation components:

D2_IS: within subpopulations

D2_ST : two loci of different gametes in a subpopulation relative to the total population D’2_IS: two loci of one gamete in a subpopulation relative to the total population

D’2_ST : total population

Ohta’s test: epistatic selection is more likely than drift if

D2_ST < D2_IS and D’2_IS < D’2_ST

N

ET

A

DAPT

:

A FUTURE TEST FOR CANDIDATE GENES

Candidate gene data can be exploited as

Genotype data Functional genomic data

Published network (at the studied or at related

model-species)

e.g. photoperiod genes

Gene expression data

e.g. Sitka spruce data

Co-expression network

No information

e.g. control genes

Annotation & idenitfying orthologues

at model species

Probabilistic functional network e.g. using AraNet

Topology of the established gene interaction network

Is the data phased? If not, use e.g. PHASE

Tests of selection

• polygenic tests (Ohta’s LD statistics) • single locus tests (FST outlier)

Statistical evidence of selection at genes and between gene-pairs

NetAdapt

• A Bayesian network-based test of selection

• Enhance our understanding of the effects of selection on gene networks • Can be used to propose new candidate genes for future studies

P

• Most traits that potentially play a role in adaptation are

controlled by many genes

• Genes do not act independently, but interact through

developmental, metabolic and biochemical networks

T

• Negligible? The elevated frequency of co-occurrence

of beneficial allele combinations at different genes is

expected to be continuously broken down by

recombi-nation.

• Not necessarily! However, if genes carrying

benefi-cial allele combinations are also functionally connected

via networks, the statistical signal of epistatic selection

may be maintained