Linkage, QTL, Mapping Populations and

Marker-Assisted-SelectionDave Douches, Allen Van Deynze and

Dave FrancisPAA SolCAP Workshop

August 2009

1

Detecting QTLs using Marker-based analysis (MBA) requires genotyping of individuals

Steps toward QTL analysis using MBA:

1- Segregating populations

2- Measure phenotypic variationThe more accurate measurement the more precise QTL identification.

3- Measure genotypic variationGenotype individualsGenetic map construction

4- Statistical analysis to find QTLs and QTL validation

Common Population StructuresF2;F3;F4;Recombinant Inbred (RI);F1-derived, intermated recombinant inbred (IRI);Doubled haploids (DH);Backcross (BC1);Association study in the base germplasm;Near Isogenic Lines (NIL).

3

QTL mapping in an F2

4

+ from Parent 2+ from Parent 1

1 2 5 6 7 8 9 103 4

…is an association study in a population where the confounding effect of pedigree has been removed.

Mapping Populations in Potato

4x crosses (between heterozygous parents)

• Mixed segregation patterns• Simplex and duplex segregation patterns

2x crosses• Testcross, F2, tri-allelic ratios• Doubled monoploid cross

– Segregation in one parent - testcross

Mappeddisease resistance

genes and QTL

Potato: Ultra-high density (UHD) genetic map

•70,344 unigenes

•150 SSRs (1000 more in ESTs)

•10,000+ AFLP markers (from 381 primer combinations)

• ordered on 136 progeny of diploid RH x SH cross

• most dense genetic map in any organism

https://gabi.rzpd.de/projects/Pomamo/Walter De Jong, Cornell University

The level of resolution needed depends on intended application

Combining QTLs between lines within a segregating population.Moving chromosome segments within heterotic pools Make germplasm wide inferences/claims about a particular chromosome segment (assuming IBD).Moving chromosome segments across heterotic pools of elite germplasm.Introgression of special trait (e.g. disease resistance) from exotic germplasm.QTL cloning

8

Incr

ease

d ne

ed fo

r res

olut

ion

Type of population depends on resolution needed

Comparison of resolution and research time for various approaches to dissect quantitative variation. The research times assume the target species has only two generations per year. NIL, near-isogenic line; RIL, recombinant inbred line

9Buckler and Thornsberry (2002)

Linkage

Linkage• Association of two or more loci on a chromosome with limited

recombination

Linkage Disequilibrium or Gametic Phase Disequilibrium• Non random association of alleles at two or more loci not

necessarily on the same chromosome• Measures co-segregation of alleles in a population• Mendel’s pea traits – showed complete linkage equilibrium and

hence independent assortment• Can arise from intermixture of populations with different gene

frequencies• Can also be produced or maintained by selection favoring one

combination of alleles over the other – e.g. selection for yield in a breeding population

Falconer and McKay (1996)10

Linkage disequilibrium

11

Linkage Disequilibrium Decay

12

r = Recombination Rate between two locir = 0.5 = Two loci are unlinkedr = 0 = Two loci are completely linked and do not independently assortFalconer and Mackay (1996)

What are QTL?

Genes are regions of DNA controlling a trait or phenotype

Quantitative Trait Loci (QTL) are regions of DNA associated with quantitative traits

• They can be defined using molecular markers that highlight specific DNA segments or genes

This can change how we manipulate these traits in breeding

A trait is quantitative/polygenic if its inheritance is controlled by two or more genes and their interaction with environment

1- A trait having a continuous distribution

2- Controlled by the interaction of many genes

Cont…

A trait is quantitative/polygenic if its inheritance is controlled by two or more genes and the their interaction with environment

Examples• Yield• Lycopene content in tomato• Early Blight (EB) resistance

3- Influenced by the environment

Trait Distribution in the Progeny

If we could observe directly the QTL we could see the 3 underlying trait distributions

Trait distribution in the F2 progeny

Modeling a Marker Locus and a Linked QTL

Single Marker Analysis modeling, assume:• One QTL locus A, • One Marker locus M,

Measure of association between A and M• Self pollinated crops:

compare AA to aa (no dominance effect)

• Cross Pollinated crop: compare AAT to aAT (dominance and additive effects are

confounded)

17

Assigned values of the genotypes at A and M

Let δ=d, d/2, and 0 for cloned, selfed, and tescrossed progenies, respectively.

18

Genotyping PhenotypingIndividual

GenotypedCloned Progeny

Selfed Progeny

Testcrossed Progeny

AA AA => a AA => a AT => aAa Aa => d 1AA:2Aa:1aa => d/2 ½ AT+ ½ aT => 0

aa aa => -a aa => -a aT => -a

Need to collect good data: Good phenotypes are important

ripeningin

fect

ion

Ann Powell UC Davis

What information do we need?

Information pertinent to the association between marker genotypes and phenotypic values.• Total number of individuals genotyped, N (and

number in each class, nMM, nMm, and nmm)• Values of a and δ.• Values of residual genetic variance and error

variance

20

Detection of a QTL

Detection of linkage between A and M, will occur only if both are segregating in the population and if they are physically linked;We cannot extrapolate results from one mapping population to the rest of our germplasm without considering disequilibrium.

21

Do we have disequilibrium between a linked marker and a QTL?

We cannot really assess very well the extent of gametic phase disequilibrium between loci in the population without very extensive mapping studies.

With SNPs we can look at marker-marker associations for a much smaller cost: we just need to genotype our germplasm.

22

The marker density needed depends on the population

23

( )

0.0

0.5

1.0

1.5

2.0

2.5

0 20 40 60 80 100

Distance of Marker from QTL (in cM)

(MM

-mm

)/a *

DH or F2 or 2xBC1F3F4IRI (n# gen RM=0)IRI (n# gen RM=2)IRI (n# gen RM=5)IRI (n# gen RM=10)IRI (n# gen RM=25)

Pow

er o

f det

ectin

g a

QTL

34.7 cM

2.4 cM

TG670

SSR10511CT6213SSR26614SSR192 TG12515SSR95 CT2012721

SSR31634CT14935CT20134I CT10725I41CT20268I42SSR13446CT10975I51TG27354CT10030I.258CT2011659LEOH106 TG5960CT10629 CT10811CT10945 LEVCOH1265LEVCOH1166CT19169SSR973SSR308 TG46581LEOH22286SSR42 TG260TG24587SSR3796CT1025998TG255103SSR65 SSR582SSR288 TG580113CT10126I115

Chr.1

CT105350LEOH3423LEOH342n7TG60812CT20522CT10682I29CT1019030TG16531SSR6633SSR9640CT1064942CT1092344TG1446SSR5 SSR60547CT10771 CT1015348CT1080151CT10279I55CT24457SSR3259LEOH34860SSR59861SSR2664TG46968LEOH11370TG64572TG33777TG53784TG16790LEOH319 LEOH17491TG15192TG154100

Chr.2

TG214 SSR6010SSR14 LEOH1271SSR3202CT10690I3CT20050 CT10772I4CT106786CT10042I13CT1045017CT8519CT10437I22TG24628LEOH11031LEOH18533TG12934CT10689I36CT10480I38SSR11144CD5145CT20195 CT10402I46CT20037 CT1043747CT10736 CT8250CT2002351LEOH22354CT1050655TG52059CT14169TG13B77TG13086TG11492

Chr.3

CT109520

SSR29610

TG1522CT10255I SSR4325LEOH36126SSR431127TG48337CT2014541CT1018447CT1032253SSR31056SSR450 CT10485CT157 SSR30657CT1080959CT1021564LEOH101 CT17868CT2002870CT19474TG16377CT1088878CT10184I79CT1013684CT1055685TG50090CT5092

CT10375121

Chr.4

CT1010CT102384TG4419CT167 SSR11514CT1003615

CT9330TG9637CT1096341CT10373I TG619CT10151I42CT10765I44CT20210I45CT10526 LEOH6346TG100A47CT11855LEOH19258LEOH31667CT1059176SSR16283SSR109 TG18584

Chr.5

CT2160CT102426CT10242I7CT1018711TG59017

CT10328I27SSR12830

TG35639LEOH24341

TG36549TG25355LEOH14657LEOH20958LEOH20061LEOH11266

CT20674TG31478

Chr.6

TG3420CT200173CT5210SSR28611L21J7a18LEOH10422SSR27627CD5728

TG18338TG17441CT1013845CD5446

CT1097455

SSR4564TG2070CT1003973LEOH22174TG49980

Chr.7

CT10152I0CT103961LEOH704

TG17616LEOH12319LEOH147 CT47CT1019220CT1001521CT92 SSR32722CT1016226TG34931TG30239SSR33544SSR6350TG33051SSR3853CT10367I62

CT26570

CT6882

Chr.8

GP390

TG189

CT14323SSR68 SSR7030CT2015931CT1000436

TG29149CT1002450SSR11056LEOH14458TG55160

CT7471LEOH11772LEOH17074TG42181

SSR333 TG328100

Chr.9

CT100820CT10082I1TG122 CT166CT1067013SSR3416SSR59617CT23419CT10105I30CT10464I31SSR31835CT1136CT1055440CT20342CT10419I43CT10078I CT1070146CT10386I CT1038653

TG40377

TG23388LEOH336 TG6389SSR22391

Chr.10

CT10683I0TG4974LEOH1765

SSR8017TG50822SSR7627

CT10120 CT20244I34

TG147 CT10781CT1091547TG38451CT10737I53CT10615I59

CT2018168TG54672TG3678

TG39389CT1002790

Chr.11

TG1800

TG689

CT21126CT10953I29

TG36038SSR2041TG56547TG11153

LEOH6664LEOH30167

CT156 CT10329I78CT1077879CT10796I LEOH27580LEOH19787CT27688

Chr.12

Tomato SNP and Indel Map

Matthew Robbins, Ohio State University

Potato SNP map

Potentially 7600 markers• Candidate genes and randomly distributed

High SNP polymorphism ratePotentially over 600 markers per chromosomeWe need over 3000 random markers to saturate the 4x genome• Bradshaw et al 2008

We need less than 1000 markers to saturate the 2x genome

Sample (population) sizeσ2 = error σ2 + residual genetic σ2

“error σ2 ” and “residual genetic σ2 ” are based on overall mean for each entry.Increasing the number of field replications will reduce error σ2 but not the between-line genetic σ2.To further reduce the denominator in the t-test. We need to add more genetic entries.

Typical split plot: – Sub-plot error: plot-level error σ2

– Plot error: genetic σ2 + (plot-level error σ2)/(rep number)

26

Statistical power

27

h2=0.10

h2=0.05

Hu and Shu 2008

28

Most popular QTL analysis methods for MBA

Single-Marker Analysis (SMA)Interval Mapping (IM)

Simple Interval Mapping (SIM)MapMaker/QTLQGENER/QTLMapQTL5

Composite Interval Mapping (CIM)QTL CartographerR/QTLMapQTL5 QTL Network 2.0

Multiple Interval Mapping (MIM)WinQTL cartographer

SMA is being used less nowadays due to its disadvantages and availability of more advanced statistical methods

Advantage:Simplest approach for detecting QTLsQuick and less memory intensive

Disadvantages:May not detect loosely linked QTLsQTL effects are often underestimatedDoes not determine the distance between marker and QTLDoes not determine the direction of QTL from marker

SIM was one the most commonly used methodology in QTL mapping

Advantage:The likelihood map represents the position of the QTL at various

points of the genome

The probable position of the QTL is given by support intervals

Requires less progenies than the SMA

Disadvantages:The number of QTL cannot be resolved

The locations of the QTL are sometimes not well resolved, the exact positions of the QTL cannot be determined

The statistical power is still relatively low

CIM has at least four advantages over SMA and SIM

Advantages:By focusing on one genome region, the multidimensional search is reduced to one-dimensional searchThe resolution of QTL locations obtained is much higher than SMA and IMThere are more variables in the model, making the model more efficientMarkers can be used as boundary conditions to narrow down the most likely QTL position

Comparison of SMA, SIM and CIM for EB resistance in tomato

SMA

SIM

CIM

EB Resistant EB Susceptible

Solanum lycopersicum X S. habrochaites

~850 BC1Plants

F1 X S. lycopersicum

BC1

(EB Susceptible) (EB Resistant)

Trait-based analysis (Selective Genotyping)

Zhang et al 2003 Mol. Breed. 12: 3-19 Early blight resistance

Foolad et al 1997 Mol. Breed. 3: 269-277 Salt tolerance

QTL mapping step by step: A practical example A raw file is used for the genetic map construction

QTL mapping step by step: A practical example A genetic map is constructed

QTL mapping step by step: A practical example Phenotypic data is collected and incorporated into the file

QTL mapping step by step: A practical example A map file is then created for QTL analysis

Key points of QTL mapping

QTL’s mapped in a bi-parental cross may not be appropriate for MAS in all populations• Marker allele and trait may not be linked in all populations.• Genetic background effects may be population specific.• Original association may be spurious.• QTL detection is dependent on magnitude of the difference

between alleles and the variance within marker classes.

Adding more genotypes is more efficient than replicating the same genotypes (provided that a minimum number of environments are sampled)h2 and size of effect important Results are trait specific

39

SolCAP Mapping PopulationsGermplasm panel• Over 300 genotypes• Over 20 programs represented

– Important advanced breeding lines– Current and Historical varieties– Parents of mapping populations– Important genetic stocks

• Phenotyping at NY, WI and OR of core traits– Evaluation of specific gravity, glucose and sucrose, chip

color, skin type, shape, vine maturity, tuber number, vitamin C, internal defects, etc.

– Accessible through the SolCAP and Solanaceae Genome Network (SGN) websites.

• Association mapping• Develop linkage hypotheses• Phenotyping of other traits

SolCAP Mapping Populations

4x population• Rio Grande X Premier Russet• 200 progeny• 7600 SNPs• 3 environments

– North Carolina, Minnesota, Idaho• Identify QTL for core traits (CHO, Vit C)

Other populations from the community• Core trait validation• Other traits of regional importance

SolCAP populations

2x population• DM1-3-516 44 x 84SD22 • 120 progeny• Testcross segregation• Map SNPs• Anchor scaffolds to the genetic map

– Link mapping to PGSC• Map core traits

Other populations from the community

Approach of MAS in Tomato

•Pathway approach for candidate gene identification and introduction to metabolic pathway databases.

•Identification of polymorphisms in data-based sequences

•Approach of potato breeding for the future

Dist MarkercM Name

CT233

17.1

TG672.1LEOH36

20.0

TG1252.0CT62

24.7

CT1497.2

LEOH17*

12.8

TG2739.3

TG599.9

CT191

10.7TG465

5.8TG2601.1LEOH7

15.1

TG25510.1

TG580

Chr 1

Dist MarkercM Name

TG608

13.9

CT205

10.7TG1653.8 LEOH23

10.9TG14

18.4

CT2446.1

TG4695.2

TG645

10.6TG537

7.0TG1673.1TG151

10.2TG154

Chr 2

Dist MarkercM Name

TG15

12.4

TG483

23.6

CT157

18.5

LEOH377.7

CT1786.5

CT194

16.5

CT501.6TG500

13.6

TG1630.0LEOH10

Chr 4

Dist MarkercM Name

CT1017.9

TG4413.6CT167

18.5

CT933.0LEOH163.0TG96

12.5

TG100A

19.5

CT118

28.3

TG185

Chr 5

IL1-

1

IL4-

3IL

4-4

IL1-

2IL

1-3

Chr 3

IL2-

4

LEOH15*

IL4-

1

LEOH17*

IL5-

2

LEOH17*

Dist MarkercM Name

TG1145.2

TG130

18.5

CT141

13.4

TG520

23.5

CT829.1

CD5110.1

TG1295.7

TG2468.2

CT85

18.2

TG214

IL3-

2

LEOH17*

IL3-

1

LEOH15*

IL3-

3

LEOH17*

LEOH17*

LEOH36

LEOH10

LEOH37

QTL Trait Origin2 L, YSD S. lyc.4 YSD S. lyc.6 L, Hue ogc

7 L, Hue S. hab.11 L, Hue S. lyc.

Audrey Darrigues, Eileen Kabelka

Example: QTL for color uniformity in elite crosses

Presenter

Presentation Notes

From SGN we can also determine a map position and identify related sequences – simply by clicking on the hot link from the gene list page.

Comments on SGN databases:

•There is a wealth of information organized and available.

•Information is not always complete/up to date.

•Display is not always optimal, and several steps may be needed to go from pathway > gene > potential marker.

•Sequence data has error associated with it. eSNPs are not the same as validated markers.

•SolCAP will be asking for feedback on how best to improve the SGN database and access via the Breeders Portal

Current SGN interfaces are aimed at the molecular biologist, with searches designed to facilitate molecular discovery• Need a portal that is trait and germplasm centered

Ability to search by traits relevant to (and defined by) breeders

When a marker is identified, a protocol for use in breeding will be provided.

Search option that only yields polymorphic marker, phenotype, or QTL results from elite germplasm

Ability to search for known parents or offspring of any given genotype

Ability to generate a list of markers that are polymorphic between any two parents

Detailed tutorial/definitions of terms and traits utilized within the database

Developing “Breeder Friendly” Tools

http://www.tomatomap.net

http://sgn.cornell.edu/

Ovate

•Missing data in SGN•Limited ability to generate tables, PCR conditions sometimes incomplete, Enzyme sometimes missing, SNP not described.

•Missing data in Tomatomap.net•SNP and sequence context requires BMC genomics supplemental table , ASPE primers, GoldenGate primers.

•2007. BMC Genomics 8:465 www.biomedcentral.com/content/pdf/1471-2164-8-465.pdf

Databases

Genotyping the core collections will impact breeding.

• Potential translational approaches: 1) introgression from other populations (domesticated or wild) 2) selection for coupling phase recombinants to establish linkage

blocks of favorable alleles (e.g. disease resistance loci)3) population development designed to maximize variation w/in

market classes4) association approaches5) whole genome approaches

• Other translational strategies will emerge under other CAPs or through innovation in public research.

Additional Thoughts from Tomato:

•Marker resources are currently not sufficient for QTL discovery in bi-parental or Association Mapping populations; in the future they should be.

•The best time to use genetic markers : early generation selection

•Restructuring of breeding program to integrate markers may include:

• 1) Increasing genotypic replication (population size) at the expense of replication (consider augmented designs).• 2) Collecting objective data.

Marker Assisted Selection in Potato

Limited number of markers available• PVY, RB, GN, VW useful• Invertase, sucrose synthase and LB QTL not validated in our

germplasm• More markers and QTL to validate from Li, et al. 2008 and Bradshaw

et al. 2008Some are PCR-based• Need for marker conversion• Use of high-throughput DNA isolation methodsApply at early generation selection phase• Seedling stage• After year 1 or 2 selection Need markers linked to core traits (e.g. glucose and starch)

Address regional, individual program and emerging needs within the Solanaceae community through a small grants program.

Six key areas we intend to allocate resources to are:

– Genotyping mapping populations in the core facilities requested by the greater breeding community

– Marker conversion – developing SNP markers linked to QTL into easily assayed (e.g. CAPs or dCAPs) markers that end-users can readily apply in their own research programs

– QTL validation and MAS

– Population development to address emerging needs

– Extension or education special projects

– New directions not envisioned at the time of proposal submission.

Small Grants Program

References

Dudley, J.W. and R.J. Lambert. 1992. Maydica 37:81-87.Falconer and McKay. 1996. Introduction to quantitative genetics. 4th ed. Longman group LTD. Essex, UK. pp 464.Laurie, C.C. et al. 2004. Genetics 168:2141-2155.Liu, B.H. 1998. Statistical Genomics. Linkage, mapping and QTL analysis.CRC press LLC. FL, USA. 611p.Schön, C.C. et al. 2004. Genetics 167: 485-498Tanksley S.D. et al. 1996. Theor. Appl. Genet. 92:213-224.Walsh, B. and Lynch, M. 1998. Genetics and Analysis of Quantitative Traits. Sinauer Associates Inc. MA, USA. 980p.Winkler, C.R. et al. 2003. Genetics 164:741-745.Xiao, J. et al. 1998. Genetics 150:899-909. Kaepler, 1997. TAG 95:618-621.Frisch, et al., 1999. Crop Science 39: 1295-1301.Knapp and Bridges, 1990. Genetics 126: 769-777.Yu et al., 2006. Nature Genetics 38:203-308.Van Deynze et al., 2007. BMC Genomics 8:465www.biomedcentral.com/content/pdf/1471-2164-8-465.pdf

55