Skip to content

Quantitative and Population Genetics

Hardy–Weinberg and its extensions, polygenic inheritance, variance components and heritability, response to selection, QTL mapping and GWAS for CSIR NET.

  • 18 min read
  • Updated
On this page · 14 sections
  1. Prerequisites
  2. The Hook: Seven Shades of Red Wheat
  3. Hardy–Weinberg Equilibrium
  4. What Disturbs the Equilibrium
  5. Polygenic (Quantitative) Inheritance
  6. Partitioning Variance and Heritability
  7. Molecular Mapping of Quantitative Traits
  8. Memory Tricks
  9. Don’t Confuse
  10. Common Mistakes
  11. Practice
  12. Quick Revision Table
  13. Key Takeaways
  14. What’s Next

Prerequisites

The Hook: Seven Shades of Red Wheat


Hardy–Weinberg Equilibrium

In 1908 G. H. Hardy and Wilhelm Weinberg independently showed that, in a large randomly mating population with no evolutionary forces, allele frequencies stay constant and genotype frequencies reach fixed proportions after one generation of random mating.

For two alleles A (frequency p) and a (frequency q), p + q = 1:

$$(p + q)^2 = p^2 + 2pq + q^2 = 1$$
GenotypeAAAaaa
Frequencyp²2pqq²

Assumptions

  1. Large population (no genetic drift)
  2. Random mating (no inbreeding or assortative mating)
  3. No mutation
  4. No migration (gene flow)
  5. No selection (equal survival and fertility of genotypes)
  6. Generations non-overlapping and allele frequencies equal in the two sexes (for the one-generation result)

Calculating allele frequencies

From genotype counts (codominant alleles):

$$p = \frac{2N_{AA} + N_{Aa}}{2N}, \qquad q = 1 - p$$

From the frequency of a recessive phenotype (assuming equilibrium):

$$q = \sqrt{\text{frequency of } aa}$$

Carrier frequency of a rare recessive. If a recessive disorder affects 1 in 10,000, q = 0.01 and carriers are 2pq = 0.0198, about 1 in 50. Carriers vastly outnumber affected individuals (here about 200 : 1), so selection against affected homozygotes removes rare alleles very slowly.

Testing for equilibrium

Compute expected genotype numbers from the observed allele frequency and apply χ². Degrees of freedom = number of genotype classes − number of alleles (because allele frequencies were estimated from the data): for two alleles, 3 − 2 = 1 df.

ExampleMMMNNNTotal
Observed3804401801000
p(M) = (760 + 440)/2000 = 0.6
Expected3604801601000

χ² = 400/360 + 1600/480 + 400/160 = 1.11 + 3.33 + 2.50 = 6.94 > 3.84 (1 df): the population is not in equilibrium; there is a deficit of heterozygotes, as expected from inbreeding or from pooling subpopulations with different allele frequencies (the Wahlund effect).

Multiple alleles

For three alleles with frequencies p, q, r: \((p + q + r)^2\). For ABO (IA = p, IB = q, i = r):

PhenotypeGenotypesFrequency
Oiir²
AIA IA, IA ip² + 2pr
BIB IB, IB iq² + 2qr
ABIA IB2pq

So \(r = \sqrt{O}\), and \(p = 1 - \sqrt{O + B}\) (because O + B = (q + r)²).

X-linked genes

MalesFemales
Recessive phenotypeqq²
Carriers—2pq

If 8% of men are red–green colour-blind, q = 0.08: expected affected women 0.0064 (0.64%) and carrier women 0.147. When allele frequencies differ between the sexes, an X-linked locus does not reach equilibrium in one generation: the frequencies in males and females oscillate around the final value, the difference halving (and changing sign) every generation.

Two loci: linkage disequilibrium

Random mating alone does not put two loci into equilibrium in one generation. The linkage disequilibrium \(D = p_{AB} - p_A p_B\) decays by a factor of (1 − r) each generation, where r is the recombination fraction:

$$D_t = D_0 (1 - r)^t$$

Tightly linked loci therefore stay in disequilibrium for many generations, which is what makes association mapping possible.


What Disturbs the Equilibrium

ForceEffect on allele frequencyKey formula or point
MutationSlowly changes p and qRates are small (around 10⁻⁵–10⁻⁶ per gene per generation in classic estimates), so mutation alone is a weak force
Selection against a recessive lethalq falls slowly once rare\(q_{t} = \dfrac{q_0}{1 + t q_0}\); generations to go from q₀ to qₜ: \(t = \dfrac{1}{q_t} - \dfrac{1}{q_0}\)
Mutation–selection balanceEquilibrium between new mutations and selectionRecessive: \(\hat{q} = \sqrt{\mu/s}\); dominant: \(\hat{q} \approx \mu/s\)
Heterozygote advantage (overdominance)Stable polymorphismWith fitnesses 1 − s (AA), 1 (Aa), 1 − t (aa): \(\hat{q} = \dfrac{s}{s + t}\) (sickle-cell and malaria)
MigrationMoves q towards the migrants’ frequency\(\Delta q = m(q_m - q)\)
Genetic driftRandom change; loss of heterozygosity in small populationsHeterozygosity falls by \(1/(2N_e)\) per generation; probability that a neutral allele is fixed = its current frequency; founder effects and bottlenecks
InbreedingChanges genotype frequencies, not allele frequenciesHeterozygotes \(2pq(1 - F)\); each homozygote gains \(pqF\)
Assortative matingIncreases homozygosity for the trait-related genes—

For the MN example above, the inbreeding coefficient that would explain the heterozygote deficit is \(F = 1 - H_{obs}/H_{exp} = 1 - 440/480 \approx 0.083\).


Polygenic (Quantitative) Inheritance

Quantitative traits (height, yield, blood pressure) vary continuously because they are controlled by many genes of small, often additive effect (polygenes) together with the environment. The multiple-factor hypothesis explains them in Mendelian terms.

Additive model with n genes

Assume each “contributing” allele adds the same amount, the genes assort independently, and dominance is absent. Crossing two extreme homozygous lines:

QuantityValue
F₁Intermediate, uniform
Number of phenotypic classes in the F₂2n + 1
Fraction of the F₂ like one extreme parent\((1/4)^n\)
Class frequenciesBinomial coefficients of \((\tfrac{1}{2} + \tfrac{1}{2})^{2n}\)
Effect of each contributing allele(difference between parental means) / 2n
nExtreme classClassesF₂ distribution
11/431 : 2 : 1
21/1651 : 4 : 6 : 4 : 1
31/6471 : 6 : 15 : 20 : 15 : 6 : 1
41/25691 : 8 : 28 : 56 : 70 : 56 : 28 : 8 : 1

Features: F₁ variance and parental variances are purely environmental; F₂ variance is larger (genetic + environmental); transgressive segregation (F₂ individuals outside the parental range) appears when each parent carries some “plus” and some “minus” alleles.

Estimating the number of genes (Castle–Wright):

$$n = \frac{(\bar{P}_1 - \bar{P}_2)^2}{8\,(V_{F_2} - V_{F_1})}$$

It underestimates n when gene effects are unequal or genes are linked.

Threshold traits

Some all-or-none traits (cleft lip, pyloric stenosis, many congenital malformations) have a continuous underlying liability; the disease appears when liability exceeds a threshold. Relatives of affected people have a higher risk than the population, the risk rises with the number of affected relatives and with the severity in the proband, and, when one sex is less often affected, affected members of that sex have more affected relatives (they needed a heavier genetic load to cross the higher threshold).


Partitioning Variance and Heritability

The phenotypic value is P = G + E. If genotype and environment are independent:

$$V_P = V_G + V_E, \qquad V_G = V_A + V_D + V_I$$
ComponentMeaning
V_A (additive)Variance of breeding values; the part transmitted from parent to offspring and responsive to selection
V_D (dominance)Interaction between alleles at a locus; not transmitted predictably
V_I (epistatic)Interaction between loci
V_E (environmental)Non-genetic variation (including measurement error)
G × E interactionGenotypes respond differently to environments; adds a further term
HeritabilityDefinitionUse
Broad sense, H²\(V_G / V_P\)Degree of genetic determination; relevant for clonal crops
Narrow sense, h²\(V_A / V_P\)Predicts response to selection and parent–offspring resemblance

Estimating heritability

MethodFormula
Inbred parents and F₁ (genetically uniform)\(V_E = \tfrac{1}{3}(V_{P_1} + V_{P_2} + V_{F_1})\) (or their mean); \(V_G = V_{F_2} - V_E\); \(H^2 = V_G / V_{F_2}\)
Backcrosses (Warner)\(h^2 = \dfrac{2V_{F_2} - (V_{B_1} + V_{B_2})}{V_{F_2}}\)
Offspring on mid-parent regressionslope b = h²
Offspring on one parent regressionslope b = ½h²
Twins (Falconer)\(H^2 \approx 2(r_{MZ} - r_{DZ})\) (assumes equal shared environments and additive genetic variance, so here H² ≈ h²)
Realised heritability\(h^2 = R / S\) from a selection experiment

Response to selection: the breeder’s equation

$$R = h^2 S$$

S (selection differential) = mean of the selected parents − population mean; R (response) = mean of their offspring − population mean. Writing S = iσ_P (i = intensity of selection) gives R = i h² σ_P.

Example. Population mean 100; selected parents average 120 (S = 20); h² = 0.3 → R = 6; offspring mean 106.


Molecular Mapping of Quantitative Traits

A quantitative trait locus (QTL) is a chromosomal region containing one or more genes that affect a quantitative trait. QTLs are found by testing for statistical association between marker genotypes and phenotype in a mapping population (F₂, backcross, RIL, DH; see Genetic Mapping).

graph LR
    A["Cross divergent parents;
build F2 / BC / RIL / DH"] --> B["Genotype markers
across the genome"] B --> C["Phenotype every line
(replicated for RIL/DH)"] C --> D["Test marker–trait
association at each position"] D --> E["LOD profile;
threshold by permutation"] E --> F["QTL position, effect,
% variance explained"] F --> G["Fine mapping with NILs;
candidate gene;
marker-assisted selection"]
MethodIdea
Single-marker analysisCompare trait means of marker genotype classes (t-test, ANOVA, regression). Simple; confounds QTL effect with distance from the marker
Interval mapping (Lander and Botstein)Scans positions between flanking markers using a likelihood model; gives a LOD profile along each chromosome
Composite interval mappingInterval mapping with other markers as covariates to absorb the effects of QTLs elsewhere
SignificanceGenome-wide LOD thresholds set by permutation tests (reshuffling phenotypes)

Estimating QTL effects in an F₂. With trait means \(\bar{x}_{MM}\), \(\bar{x}_{Mm}\), \(\bar{x}_{mm}\) for the marker genotypes:

$$a = \frac{\bar{x}_{MM} - \bar{x}_{mm}}{2}, \qquad d = \bar{x}_{Mm} - \frac{\bar{x}_{MM} + \bar{x}_{mm}}{2}$$

d ≈ 0 means additive gene action; d ≈ ±a means complete dominance; |d| > |a| suggests overdominance. The proportion of phenotypic variance explained (R²) measures the QTL’s importance.

Association mapping and GWAS

Genome-wide association studies test hundreds of thousands to millions of SNPs in unrelated individuals for association with a trait or disease, relying on linkage disequilibrium between the genotyped SNP and the causal variant.

  • Results are shown as a Manhattan plot (−log₁₀ P against genomic position).
  • Because so many tests are made, the conventional genome-wide significance threshold in human GWAS is P < 5 × 10⁻⁸.
  • Population structure can cause false associations; it is corrected with principal components or mixed models.
  • Most associated variants have small effects and together explain less of the heritability than twin studies predict (missing heritability).
  • Linkage vs association: linkage mapping (families, mapping populations) has low resolution but works with few markers; association uses historical recombination, so resolution is high but many markers and large samples are needed.

Uses in breeding

ApproachWhat it does
Marker-assisted selection (MAS)Select plants carrying the favourable allele of a marker tightly linked to a gene or QTL, without phenotyping
Marker-assisted backcrossingForeground selection for the target gene + background selection for the recurrent parent genome, recovering it in fewer generations
Gene pyramidingCombining several resistance genes in one line (hard to do by phenotype alone, since the phenotype of one gene masks another)
Genomic selectionPredicting breeding values from genome-wide markers without first identifying QTLs

Memory Tricks

Don’t Confuse

ThisvsThat
Inbreeding: changes genotype frequencies onlyvsSelection, drift, mutation, migration: change allele frequencies
H² (all genetic variance)vsh² (additive only)
Selection differential S (parents vs population)vsResponse R (offspring vs population)
Polygenic (many genes → one trait)vsPleiotropy (one gene → many traits)
Continuous traitvsThreshold trait (discrete phenotype, continuous liability)
Linkage disequilibrium (non-random allele association across loci)vsLinkage (physical proximity; loci can be linked and in equilibrium)
QTL mapping (designed cross, few meioses)vsGWAS (unrelated individuals, historical recombination)
HWE χ²: df = genotypes − allelesvsMendelian ratio χ²: df = classes − 1

Common Mistakes


Practice

All questions below are original practice questions written in CSIR NET style. They are not previous-year questions.

Level 1: Foundation

  1. An autosomal recessive disease affects 1 in 10,000 people. What is the carrier frequency?
  2. Which variance component is used in narrow-sense heritability?
  3. How many phenotypic classes appear in the F₂ of a cross differing at four additive genes?
Answers
  1. q = 0.01; 2pq = 0.0198, about 1 in 50.
  2. Additive genetic variance (V_A).
  3. 2n + 1 = 9.

Level 2: Part B Style

Q1 (practice). In a population at Hardy–Weinberg equilibrium, 8% of males show an X-linked recessive trait. The frequency of heterozygous (carrier) females is about:

  • (a) 0.0064
  • (b) 0.08
  • (c) 0.147
  • (d) 0.16
Answer and explanation

(c). q = 0.08 (male frequency). Carrier females = 2pq = 2 × 0.92 × 0.08 = 0.147. (a) is affected females (q²); (d) is the approximation 2q, which overestimates when q is not small.

Q2 (practice). A recessive lethal allele has frequency 0.1. Assuming complete selection against homozygotes and no new mutation, how many generations are needed to reduce its frequency to 0.01?

  • (a) 9
  • (b) 10
  • (c) 90
  • (d) 100
Answer and explanation

(c). t = 1/q_t − 1/q₀ = 100 − 10 = 90 generations.

Q3 (practice). The mean height of a plant population is 100 cm. Plants averaging 120 cm are selected as parents, and their offspring average 106 cm. The realised heritability is:

  • (a) 0.06
  • (b) 0.30
  • (c) 0.53
  • (d) 0.88
Answer and explanation

(b). S = 20, R = 6, h² = R/S = 0.30.

Q4 (practice). Correlations for a trait are 0.80 between identical twins and 0.50 between fraternal twins reared together. Falconer’s heritability estimate is:

  • (a) 0.30
  • (b) 0.50
  • (c) 0.60
  • (d) 0.80
Answer and explanation

(c). H² = 2(0.80 − 0.50) = 0.60. The estimate assumes MZ and DZ twins share environments equally and that genetic variance is additive (dominance would inflate it).

Level 3: Part C Style

Q1 (practice). Two inbred lines of a crop and their progenies give the following variances for grain weight (g²): P₁ = 4, P₂ = 4, F₁ = 4, F₂ = 16, B₁ (F₁ × P₁) = 12, B₂ (F₁ × P₂) = 14.

(i) Estimate V_E and V_G in the F₂, and the broad-sense heritability. (ii) Estimate the narrow-sense heritability by Warner’s method. (iii) What do (i) and (ii) together tell you about gene action?

Answer and worked solution

(i) P₁, P₂ and F₁ are genetically uniform, so their variance is environmental: V_E = 4. V_G = 16 − 4 = 12; H² = 12/16 = 0.75.

(ii) \(h^2 = \dfrac{2(16) - (12 + 14)}{16} = \dfrac{6}{16} = \mathbf{0.375}\). So V_A = 6.

(iii) V_A = 6 of V_G = 12: the other half is non-additive (dominance and interaction), V_D ≈ 6. Half the genetic variance will not respond to selection; the population would gain more from exploiting dominance (hybrids) than from simple mass selection.

Q2 (practice). In a sample of 1,000 people typed for the MN blood group: MM 380, MN 440, NN 180. Which statements are correct?

(P) The frequency of the M allele is 0.6. (Q) The χ² test with 2 degrees of freedom shows the population is in Hardy–Weinberg equilibrium at P = 0.05. (R) There is a deficit of heterozygotes, consistent with an inbreeding coefficient of about 0.08. (S) Inbreeding of this degree would also change the allele frequency of M in the next generation.

  • (a) P and R only
  • (b) P, Q and R
  • (c) P, R and S
  • (d) Q and S only
Answer and worked solution

(a) P and R only.

  • p(M) = (2 × 380 + 440)/2000 = 0.6 (P correct).
  • Expected 360 : 480 : 160; χ² = 1.11 + 3.33 + 2.50 = 6.94. Degrees of freedom = 3 genotypes − 2 alleles = 1 (critical 3.84), so the population departs from HWE. (Q) is wrong on both the df and the conclusion (even with 2 df, 6.94 exceeds 5.99).
  • F = 1 − 440/480 = 0.083 (R correct).
  • Inbreeding redistributes genotypes but does not change allele frequencies (S wrong).

Q3 (practice). Two inbred maize lines differ in ear length: P₁ mean 10 cm, P₂ mean 26 cm. The F₁ is uniform at about 18 cm. Among 2,048 F₂ plants, about 8 are as short as P₁ and about 8 as long as P₂. A marker linked to one QTL gives F₂ class means MM 20.0 cm, Mm 18.0 cm, mm 16.0 cm.

(i) How many genes appear to control ear length, and how much does each contributing allele add? (ii) How many phenotypic classes would you expect in the F₂ without environmental variation? (iii) Estimate the additive and dominance effects at the marked QTL and describe gene action there.

Answer and worked solution

(i) 8/2048 = 1/256 = \((1/4)^4\) → n = 4 genes. Range 26 − 10 = 16 cm spread over 2n = 8 contributing alleles → 2 cm per allele.

(ii) 2n + 1 = 9 classes (10, 12, 14, …, 26 cm).

(iii) a = (20.0 − 16.0)/2 = 2.0 cm; d = 18.0 − (20.0 + 16.0)/2 = 0. Purely additive gene action at this QTL, consistent with the additive model in (i). (The QTL estimate a is per allele substitution, matching the 2 cm per allele, which would hold if the marker were tightly linked; looser linkage would shrink the apparent effect.)


Quick Revision Table

ItemFact
HWEp² + 2pq + q² = 1; one generation of random mating; Hardy and Weinberg 1908
AssumptionsLarge, random mating, no mutation, migration or selection
Recessive phenotypeq = √(freq aa); rare-allele carriers ≈ 2q
HWE χ²df = genotypes − alleles (1 for two alleles)
ABOr = √O; p = 1 − √(O + B)
X-linkedMales q, females q²; oscillating approach to equilibrium
LD decay\(D_t = D_0(1 - r)^t\)
InbreedingHeterozygotes 2pq(1 − F); allele frequencies unchanged; F = 1 − H_obs/H_exp
Selection vs recessive lethal\(t = 1/q_t - 1/q_0\)
Mutation–selectionRecessive \(\sqrt{\mu/s}\); dominant μ/s
Overdominance\(\hat{q} = s/(s + t)\)
DriftLoss of heterozygosity 1/(2N_e) per generation; fixation probability = frequency
Polygenes2n + 1 classes; extreme (1/4)ⁿ; Nilsson-Ehle wheat (3 genes, 1/64)
Castle–Wright\(n = (\bar{P}_1 - \bar{P}_2)^2 / 8(V_{F_2} - V_{F_1})\)
VarianceV_P = V_G + V_E; V_G = V_A + V_D + V_I
H² / h²V_G/V_P / V_A/V_P
EstimatingV_E from P₁, P₂, F₁; Warner h² = [2V_F₂ − (V_B₁ + V_B₂)]/V_F₂
RegressionMid-parent b = h²; single parent b = ½h²
TwinsH² ≈ 2(r_MZ − r_DZ)
Breeder’s equationR = h²S = i h² σ_P
QTLSingle marker, interval (Lander–Botstein), composite interval mapping; permutation thresholds
QTL effectsa = (MM − mm)/2; d = Mm − mid-point
GWASLD-based; Manhattan plot; P < 5 × 10⁻⁸; correct for structure
BreedingMAS, marker-assisted backcrossing, pyramiding, genomic selection

Key Takeaways

What’s Next

Advertisement
Advertisement