Quantitative Genetics Flashcards
What are complex traits?
Measures of an individual that vary in degree rather than kind
Eg. growth rate, height, weight
Study of the genetics known as quantitative genetics
Why are complex traits important?
Evolution - much adaptation results from natural selection operating on quantitative traits, Eg. beak size in Darwin’s finches
Animal & plant breeding - target for artificial selection in domesticated animals and crops
Human medicine - many health problems results from diseases whose susceptibility varies on a continuous scale, Eg. BMI predicts cancer risk
What are the fundamental properties of complex traits?
Resemblance between relatives, share alleles inherited identical by descent, may share same genotype at some loci, tend to be exposed to similar environments
Respond to selection, artificial and natural, offspring from selected population tend to score higher than offspring of an unselected population, means changes across generations
Show inbreeding depression and hybrid vigour, inbreeding reduces fitness of offspring which can be restored by crossing with unrelated strain (hybrid vigour), changes frequency of homozygotes
Many show a continuous distribution
What are the causes of variation for complex traits?
Genetic - alleles at many loci affecting the trait segregate simultaneously, each allele at each locus affecting the trait generally has small effect, infinitesimal model
Environmental - increase/decrease trait, blur distinctions between genotypes
P = G + E
How can continuous variation arise from many loci?
Assume several loci affect a trait each with two alleles
At each locus each alleles of type 1 adds one unit to trait value
Heterozygote is exactly intermediate between homozygotes - additive gene action
Assume allele frequencies are same for each locus and genotypes segregate in HW proportions
I there’s 1 locus each genotypes has unique phenotype, only 3 phenotypes, if 2 loci 9 genotypes and 5 phenotypes, intermediate phenotypes generated by several genotypes
With increasing loci becomes more like a normal distribution
What are the measurable properties of complex traits?
If we directly measure an individual for a trait we obtain its phenotypic value
If we measure a sample of individuals in a population can estimate some statistical properties of the quantitative trait - mean, variance, covarience
What is variance?
A measure of the dispersion of the distribution of values
What is covariance?
A measure of the association between pairs of values
How do you calculate variance?
(sum of (difference of each value)^2) / n-1
Divide by n-1 as you always overestimate variance in a sample of population
How do you calculate covariance?
(sum of (difference of each X value * difference of each Y value)) / n-1
How can genetic variation be quantified?
The amount of phenotypic resemblance among relatives for a quantitative trait can be used to quantify the amount of genetic variation for the trait
How can phenotypic resemblance between relatives be measured?
The degree of similarity among a group of relatives for the trait compared to random members of a population (covariance of family members)
Equivalent to
Extent of differences in phenotype between different families (variance between families)
What is a calculation to quantify resemblance between relatives?
P = G + E
P - phenotypic value
G - genotypic value
E - environmental value
What do we assume for the environmental value (E)?
Usually assumed not to be transmitted to offspring
Mean assumed to be zero
How can we estimate the genotypic value (G)?
If you raise multiple cloned progeny from a parent in random environments, G, of the parent estimated as the mean phenotypic value (P) of the progeny BUT not a likely scenario
How do we model the components of the genotypic value (G)?
G = A + D
G - genotypic value
A - breeding value, transmitted from parent to offspring, individual judged by the average value of its offspring
D - dominance deviance, interactions between pairs of alleles not transmitted from parent to offspring
G is not wholly transmitted from parents to offspring as alleles and not genotypes are transmitted across generations
How can we measure breeding value (A) of an individual?
Mate one male to random unmated females and raise offspring in multiple, random environments
Difference between mean offspring’s phenotypic value and population mean in half male parent’s breeding value (A/2)
What is heritability of a trait?
The proportion of the phenotypic variation for a quantitative trait that’s genetic
What can we predict from the heritability?
Resemblance between relatives
Response to selection
What are the 2 measures of heritability?
Broad sense heritability (H^2): for clonal organisms, proportion of phenotypic variance caused by variation among individuals in their genotypic values (G)
H^2 = Vg/Vp
Narrow sense heritability (h^2): diploid organisms, proportion of phenotypic variance caused by variation among individuals in their breeding values, A:
h^2 = Va/Vp
Which measure of heritability is more important?
Narrow sense heritability
Directly determines parent-offspring resemblance and response to selection
So need to estimate Va to estimate heritability
Vp easy to estimate as it’s the variance of phenotypic values
How is the variance in breeding values (Va) calculated?
Mate male parents to multiple, randomly chosen females from a population to generate several large half-sib families
Dominance and environmental effects assumed not to be transmitted
Do this for multiple males and find the variance
Variance of these family deviations from the population mean = Vhs = Va/4
What does heritability not indicate?
Doesn’t indicate absolute amount - can only be compared across traits, scales, populations
Doesn’t indicate number of genes
Low heritability doesn’t mean not genetically determined, just has a low genetic variance
What are the 2 steps to finding heritability interference?
Determine arithmetic relationship between heritability and covariance between particular relatives we are interested in
Work out a way to statistically estimate covariance
How do we estimate heritability from parent-offspring resemblance?
Covariance = Va/2
Share half alleles identical by descent
Estimated as twice the slope of regression of offspring phenotype on parent phenotype
h^2 = 2bop = Va/Vp
What assumptions are made when heritability is estimated from parent-offspring resemblance?
Linear relationship between parent and offspring phenotypic values
Absence of environmental sources of resemblance between parents and offspring
Fertility and viability aren’t correlated to the trait under study
How do we estimate heritability from full-sib resemblance?
Covariance = Va/2 + Vd/4
Different to parent-offspring as full-sibs share same genotypes at a locus 25% of the time
Use ANOVA to estimate Vb
What is the intraclass correlation of full sibs (tFS)?
tFS = Vb/Vp = Vb/(Vb+Vw)
Get an upwardly biased estimate of h^2 by doubling tFS - contains some dominance variance
What impacts heritability estimation in full-sibs?
Maternal effects common and impact the estimate so most people don’t use it
How do we estimate heritability from half-sib resemblance?
Most reliable, especially paternal half-sibs (only share a father) shouldn’t share environment and only share 1 allele
Mate single male to multiple females, for multiple males
Covariance = Va/4
What is artificial selection?
Deliberate choice of select group of individuals (usually phenotypically superior) for breeding
Results in change in mean phenotype over generations
What is an example of artificial selection?
Mean yield of corn in the USA
Nearly 5-fold increase since 1930
About 50% due to genetic change from artificial selection
How do we describe the response to artificial selection?
Assume from a population top x% are parents of next generation Mp = mean of complete population Ms = mean of selected group of parents S = Ms - Mp selection differential Mo = mean offspring R = Mo - Mp response to selection
Why is Mo > Mp usually?
Some of the selected parents have favourable genotypes and pass favourable alleles to their offspring
Why is Mo < Ms usually?
Some selected parents don’t have favourable genotypes; their good phenotypes result from the environment
Alleles, not genotypes are transmitted to offspring, and good genotypes are disrupted by Mendelian segregation and recombination
How do we predict response to selection
Need to define relationship between R and S
Points are plotted representing pairs of offspring and mid-parental phenotypic values, both expressed as deviations from mean
R and S both lie on the regression line R = b*S
b = slope of offspring mid-parent regression (heritability)
What is the calculation for response to selection?
R = h^2*S
What assumptions do we make when calculating response to selection?
Assume a linear relationship between offspring and mid-parent
Absence of environmental causes of resemblance between offspring and mid-parent
Absence of viability (natural) selection on progeny/fertility (natural) selection in parents
What limits response?
There’s a limit to how large S (selection) can be imposed by the phenotypic variance (Vp) of the character
If character fairly invariant (Vp small), S can’t be very large
S often measured in standard deviation units as selection intensity
Why does selection response vary from generation to generation?
Random drift due to small population size
Error in estimation of the mean at each generation
Variation of selection differentials due to fertility/viability differences
Why is response to selection in opposite directions often asymmetric?
Random drift
Selection differential higher in one direction than the other due to fertility/viability differences
Inbreeding depression
Extreme allele frequencies
Why do some selection responses show selection limits?
Exhaustion of variation - no longer any variance that will increase or decrease the trait
Due to variation present but not leading to a response - eg. dominance involved
What makes a substantial contribution to long term selection responses?
New mutational variation - demonstrated by selecting in inbred lines
What causes traits to not dependent?
Pleiotropy - same loci affect both X and Y
Linkage disequilibrium - loci affecting X are associated with locus affecting Y
What is ‘realised’ heritability?
Empirical measurement of the effectiveness of selection
Allows us to estimate heritability from a selection experiment
h^2 = R/S (selection differential)
h^2 estimated from regression of sum of R on sum of S
What are some examples of artificial selection affecting natural populations?
Fishing - putting back smaller fish selects for smaller fish
Trophy hunting - selects against larger horns
Can we predict evolution by natural selection using
R = h^2 x S?
With long-term monitoring can: Measure heritability Measure selection Measure change between generations Not so accurate
Why is predicting evolution of natural populations difficult?
Environment isn’t constant - may change in opposite direction masking evolution
Estimates of heritability are wrong/too simplistic - environment interactions, changes across environments
Selection on correlated traits
What is a QTL?
2 definitions:
A region of the genome influencing a quantitative trait
OR
A gene influencing a quantitative trait
Why study QTLs?
Need to ID genes responsible for genetic variation and adaptation in natural populations
ID alleles causing predisposition to common multifactorial diseases in humans - prevention and diagnosis
Improve efficiency of selective breeding using marker-assisted selection and genomic selection
Introgress exotic genes into commercial breeds eg. disease resistance by marker-assisted introgression
What are the goals of QTL mapping?
ID all loci affecting the trait down to the limit of experimental resolution
ID the genetic location of the loci on linkage maps
ID the genes themselves
What are the requirements for QTL mapping?
Needs genetic markers - microsatellites/SNPs
Non-random associations between alleles at QTLs and alleles linked at marker loci
Recombination breaks down non-random associations
How does QTL mapping using crosses between lines work?
Assume we have 2 lines that have diff mean values for quantitative trait
Fixed for diff alleles at QTL that affects that trait and several marker loci that don’t
Markers can be observed for genotyping but the QTL can’t be directly observed
‘Mapping population’ produced from crosses between lines - backcrossing
Population genotyped at markers and phenotypic values for trait recorded
Average phenotypic values of diff marker genotypes compared
Presence of QTL indicated by significant difference in mean phenotype between marker genotype classes
What indicates the presence of a QTL?
Significant difference in mean phenotype between marker genotype classes
How is the effect of QTL on a trait estimated?
True effect QTL: a
Recombination fraction between marker and QTL: c
`a = a(1-2c)
If c = 0 (complete linkage) a = a
If c = 0.5 (no linkage)
a = 0
Why are QTL effects underestimated?
Markers aren’t ever completely linked to the QTL - will always be some recombination
When using multiple markers how can you test which is most closely associated with the QTL?
Differences in mean between marker genotypes - use t-test to test significance
What is interval mapping?
If we know the distance in recombination units between marker loci can more accurately estimate position of QTL
Expressed as ‘LOD score’
What are examples of QTL mapping using crosses between lines?
Usually involve markers that scan the whole genome for QTLs eg:
Between inbred strains of mice differing for alcohol addictive behaviour
Between breeds of pigs that differ for fatness and fertility
Between species that cross in the lab to study evolution of male secondary traits
Why are the number of QTLs detected a minima?
Closely linked QTLs with opposite effects tend to be missed
Linked QTLs with effects in the same direction appear as a single large effect QTL
QTL size detection limit varies with the size of experiment and the trait’s properties
What patterns are observed for the number of QTLs detected?
Often modest numbers of moderate QTLs - polygenic but not infinitesimal
Major QTLs explain high proportion of the genetic variation occasionally found
QTLs have extremely variable effects
Minor effect QTLs may not be mappable
What is association mapping?
Uses SNPs to investigate QTLs in variable populations rather than lines
Still relies on linkage disequilibrium (LD)
LD only persists over very short map distances in natural populations
QTL and marker alleles not associated unless very tight linkage has maintained LD between marker alleles and an ancestral mutation
Eg. Genome-wide association studies (GWAS) - marker alleles that cause the disease at a higher/lower freq in affected vs unaffected individuals