PSY 5137 Exam 2 Study Material on Heritability and Genetic Variance

Multifactorial/polygenic inheritance

- It is influenced by more than one gene

- Influenced by additive effects of many multiple genes, with small effects

Multifactorial/polygenic inheritance versus Mendelian inheritance

Polygenic refers to a trait that is influenced by multiple genes

Mendelian refers to a trait being influenced by a single gene

How does a quantitative phenotype differ from Mendelian conditions (from Exam 1)?

Quantitative phenotypes are on a continuum, and can vary continuously (height, weight, blood pressure, etc.)

Mendelian traits are binary (do you show the phenotype? Yes or no? ex. eye color)

Most important difference between monozygotic and dizygotic twins

Monozygotic: share all genetic info

Dizygotic: share 50% of the genetic info

How do MZ and DZ twins differ

DZ twins are ALWAYS dichorionic (have own placenta and amniotic sac)

MZ twins can be EITHER mono- or di- chorionic

Typically MZ twins 2/3 monochorionic (share both placenta and amniotic sac) and 1/3 dichorionic

Why MZ twins occur

No definitive understanding of cause

Why DZ twins occur

Two eggs and two sperm

Precise definition of heritability

Proportion of variance for a given trait that can be accounted for by genetic differences

How variance in a phenotype can be decomposed

Environmental variance: C (common/shared environment) + E (measurement error/non-shared environment)

Genetic variance:

A (additive genetic) + I (epistasis) + D (dominance)

When phenotypic variance is normed to 1, which components of genetic variance (3 components) and environmental variance (2 components) are added up to equal 1?

1.0 = (a^2 + i^2 + d^2) + (c^2 + e^2)

Broad sense heritability versus narrow sense heritability

Narrow sense heritability:

the ratio of additive genetic variance to total phenotypic variance.

Variance caused by the additive genetic factors (ACE model)

Hn^2 = a^2

Broad sense heritability: the ratio of total genetic variance to total phenotypic variance.

Includes additive factors, dominance, and epistasis

H^2 = a^2 + d^2 + i^2

Heritability is useful for

- Predicting response to selection

- Determining the extent to which genetic factors predict phenotypic differences

- Determining evidence for gene mapping effects

How heritability cannot be thought of as a fixed biological constant

Heritability is NOT a constant and can differ between groups/countries

Ex. heritability of height has increased over time

Biometric decomposition

The variance of the phenotype can be decomposed into genetic and environmental conditions

Basic terms of additive genetic variance

Environmental Variance:

C (common/shared enviro) + E (non shared + measurement error)

Genetic Variance:

A (additive genetic) + I (epistasis) + D (dominance)

Which of those 5 terms are most important in explaining the similarity of relatives? ACE model?

A (additive genetic) + C (common/shared enviro) + E (non shared + measurement error)

** There is little evidence for D and I effects

Additive genetic variance shared by MZ

100%

Additive genetic variance shared by DZ

50%

Additive genetic variance shared by parent-offspring

50%

Additive genetic variance shared by half-siblings

25%

Monochorionic (MC) and dichorionic (DC) twins

MC (monozygotic) twins: - share a single placenta

- were born at the same time

- share 100% of their genetic variance

- they will have identical appearance

- often termed as identical twins

DC (dizygotic) twins:

- do not share a placenta

- were born at the same time

- share 50% genetic variance

- will not have an identical appearance

Why are MC and DC twins a threat to the classical twin study design?

Chorionicity is the point in time which the zygote splits in half to create twins. This can be detected by an ultrasound. The threat this poses to the CTS design is because dichorionic do not have a shared environment

Why do behavioral geneticists think of the classical twin study design as a “natural experiment”

MZ twins share 100% of their genetic material and DZ twins share about 50%, which means that comparisons of the similarities can be made when MZ twins are reared together vs when DZ twins are reared together

same environment, so can compare genetic effects

For a given trait, if MZ twins are more similar than DZ twins, what does this imply about the genetic or common environmental sources of variance for that trait?

What if MZ twins and DZ twins are equally similar?

1. For a given trait, if MZ twins are more similar than DZ twins, heritability estimates are likely pretty high for that trait, and genetics would likely be what is causing individual differences

2. If MZ twins and DZ twins are equally similar, then the environments in which they grow up in will likely be what causes individual differences

Two major assumptions of the classic twin study design

1. MZ twins are representative of the population both physically and psychologically

2. MZ twins and DZ twins have a very similar shared environment

What an we say about whether research suggests these assumptions pose a threat to the validity of results from twin studies?

1. MIDUS determined that MZ twins and DZ twins are not perfectly similar in a shared environment, but differences are negligible and likely have little effect to validity

2. Rates of mental illness hover around 1 percent in twins, which is not significantly larger than the general population

How biometric estimates of variance can be estimated from dizygotic and monozygotic twin correlations

(Falconer's equations, ACE model)

The contribution of the additive genetic component (a2) is proportional to the "proportion of segregating genes in common"

This value is 100% for MZ twins, 50% for DZ/1st degree, etc.

The shared (common, c2) environment component contributes to the similarity of reared-together relatives

How to calculate heritability (a2) and shared/common environment (c2) from MZ and DZ twin correlations

a2=2(rMZ-rDZ)

c2=2rDZ-rMZ

e.g., Estimating heritability of height

MZ Twins: r=0.91

DZ Twins: r=0.53

a2 = 2(0.91 - 0.53)

a2 (heritability) = 0.76

Basic findings of the Polderman et al. (2015) meta-analysis of twin studies:

Did they find much evidence for non-additive genetic effects?

What general conclusions did they reach about the influence of the shared/common environment?

Results provided evidence that all human traits are heritable. Not a single trait had a weighted heritability estimate of zero.

Basic logic of an adoption study

Provides the ability to observe how genetic factors shape behavior when influenced by separate environmental factors.

Studies genetically similar subjects in reared separate environments.

How adoption studies aid in our understanding of environmental sources of variance

Researchers can determine whether the variance is environment or genetic influenced.

Confound of the same rearing environment is eliminated.

Basic limitations and assumptions of adoption studies

- All adopted households are the same and representative of the population

- Adopted houses usually are much better off on average than the general population

These assumptions are not true and can lead to missing out on some environmental causes of behavior.

Is there a problem of selective placement?

Is there evidence that selective placement undermines the validity of adoption study results?

- The placing of adopted children in homes resembling those of their biological parents in social and educational terms

- Selective placement has been found to occur at roughly 11% meaning that it has a small, but negligible effect on outcomes of adopted children

- Selective placement does not greatly undermine adoption studies

Minnesota Study of Twins Reared Apart (MISTRA)

Research project by Thomas J. Buchard Jr. in 1979

Unique and unusual type of twin MISTRA studied?

MISTRA was unique from both adoption studies and twin studies in that the methodology featured a combination of both

The sample consisted of twins reared apart ex. twins adopted into different families

MISTRA major conclusions

1. All human behavioral traits are heritable (40-50%)

2. The shared family environment has minimal impact on individual differences in behavior (0-10%)

3. The non-shared environment exerts a major influence on individual differences in behavior (50%)

How can adoption studies be used as a quasi-intervention

Quasi-intervention is when you evaluate interventions but do not use randomization

In adoption studies, typically the average adoptive homes are advantaged since they have to go through many hurdles in order to be considered eligible to adopt. Therefore, when you study adopted individuals you can also evaluate the impact of having advantaged rearing

What have adoption studies shown about the influence of the common/shared rearing environment for traits like personality?

What types of traits might be an exception to this general pattern?

Adoption studies generally have little evidence showing that the shared environment has an affect on personality

Certain traits such as religiosity tend to be exceptions for these patterns

Gene-environment interaction

When both genetic and environmental factors combine or interact, to influence individual differences

How might a gene-environment interaction look on a graph?

This may look like a cross between two lines

Example of a specific gene environment interaction we discussed in class?

Depression is an example of GxE interaction, because it is heritable but also can be triggered by life stresses

Understand how the diathesis-stress model relates to the concept of gene-environment interaction

The diathesis-stress model explains a disorder or its risk by using the diathesis (predisposition for the disorder) and its effect on environmental factors

Basic function of the MAO-A enzyme

The MAO-A enzyme is a deamination enzyme which deconstructs amine (peptide) based hormones

This includes neurotransmitters such as Dopamine, Serotonin and Epinephrine

Caspi et al. (2002) findings about the relationship between MAO-A alleles, childhood maltreatment, and antisocial behavior?

What limitations does the Caspi et al. study have?

Capsi et al found links between having variants of MAO-A (only those that decrease MAO-A activity/substrate binding), and an individual's propensity to exhibit anti-social behaviors, mediated by childhood maltreatment.

Some issues exist with this study however, particularly with sample size.

Gene-environment correlation

The phenomenon in which the individual's genetics influence their self-created environment (or the parental-created environment)

Passive, reactive (or evocative), and active gene-environment correlation?

What might be an example of each?

Passive

The environment is created by a parental figure due to shared genetics

Reactive

An individual's behavior provokes an environmental state which is corresponding to one's genetic makeup

Active

One's genetics actively influences one's choice to create/choose an environment that corresponds to one's genotype

Environmental variables that would contribute to the variance in common or shared environments

- Parental influence

- Diet

Environmental variables that would contribute to the variance in a unique or non-shared environment

- School or work environment

- Peer groups

Which of the two environmental variance components in the ACE decomposition is typically larger?

The unshared environmental traits are more influential to someone

Impact of being reared together or apart on the correlation between the personality scores of relatives?

Which kinship type shows a higher correlation—monozygotic or dizygotic twins?

What about adoptive siblings?

There doesn't appear to be very much correlation on personality when it comes to being reared together vs. reared apart

Monozygotic twins tend to show a bit of a higher correlation on personality traits compared to dizygotic and adoptive siblings

Wilson effect

The shared environment influences on general cognitive ability decline over time while the genetic influences increase over time

In early life, passive-gene environment correlation dominates as parents have a large effect on early life environments. But with age, active-gene environment correlation dominates and parents have little influence on that

C^2 declines as A^2 increases with age

How does the heritability of IQ change from childhood to adulthood?

The heritability of IQ increases with age.

How does the influence of a shared family environment on IQ change from childhood to adulthood?

Shared family environment influence on IQ decreases with age

Traits that show evidence of changing variance components over time

Intelligence or IQ shows that initially shared environment when younger plays a major role, but ultimately genetics is more influential with age

Traits that show strong persisting resemblance between adoptive relatives

Three laws of behavior genetics

1. All human behavior traits are heritable.

2. The shared family environment has little impact on differences in behaviors

3. Most individual differences in behavior are due to the non shared environment

Examples of exceptions to these three laws

Positional cloning

A method of gene identification where a gene for a specific phenotype is identified by its approximate location on the chromosome

Identifying a candidate region.

NOT learning anything about the function of the gene by doing this, rather the location of the disease associated gene on a chromosome.

This strategy can be used even without knowing its function in the disease.

Is positional cloning able to tell us anything about specific genetic variants or their causes?

No, it is not able to find causal variants

Allelic association

The comparison between an affected group or control group and allele frequencies

In allelic association, what statistical metric is most commonly used to represent risk?

The odds ratio (OR)

Odds ratio of 2+ is VERY strong

Candidate gene approach to identifying genetic variants?

Strengths?

Weaknesses?

Strengths

- Can test for smaller effects by increasing sample size

- Can test for involvement of a gene in the causal pathway to disease

Weaknesses

- Prone to false positive results, publication bias, mismatching cases and controls

- Need to have the candidate genes

Explain what ultimately came of the candidate gene approach to identifying genetic variants associated with quantitative phenotypes

It failed

The right candidates were investigated, but the study was underpowered (that is, adaquetly powered to idenitify risk varaiants with odds ratios above 1.4)

The wrong candidated were investigated, even tho the candidates emerged from a systematic and replicated process, they may have been false positives

SNP imputation or tagging

A tag SNP is an SNP in a region of the genome with high linkage disequilibrium that represents a group of SNPS

A SNP that is close to another SNP can be tagged because it is highly correlated

If a SNP is far away, it is not likely to be tagged

Based on this strategy, you can identify genetic variation without genotyping every single SNP

In the Sanders et al. (2008) candidate gene study of schizophrenia, they found 30 genetic variants significant at p < .05 and 3 variants significant at p < .01. Given the total number of independent tests the authors conducted, do these variants actually reach genome-wide significance? Why or why not?

30 of the chi-square tests significant (by chance) with a p<0.05 and 3 significant (by chance) with p<0.01

They did not find anything statistically significant

Linkage disequilibrium (LD)

Non-random, population level association between alleles at a linked loci

Can lead to population level (indirect) associations between non-functional genetic markers and phenotypes

How does imputation/tagging take advantage of linkage disequilibrium?

The existence of linkage dis. implies we do not need to genotype the casual variant (D) but rather just alleles at loci near the casual locus

Define haplotype

How is it relevant to LD?

Alleles at different loci inherited together on same chromosome

Over time, showed that recombination will diminish the haplotype association

How are meiosis/recombination, number of generations, and physical distance in the genome relevant to the concept of LD?

Alleles at linked loci will erode over time (with meiosis over generations) due to recombination

Likelihood of recombination over generations is proportional to how far apart the loci are

Lower chance between loci closer physically

Genome-wide association study (GWAS)

GWAS identifies genes associated with a particular disease or trait

The approach compares many different genomes from many different people to find genetic markers associated with a disease

Is GWAS designed to detect common or rare genetic variants?

GWAS is designed to detect common variants (frequency greater than 0.01)

GWAS would be underpowered for rare variants. You would need a massive sample to find cases of a rare variant.

Common variants exist more in the general population. Common and likely ancient variants means LD will extend over thousands of bases

How does sample size relate to the number of genetic variant "hits" found across GWAS?

How does sample size relate to the total amount of variance explained in the phenotype?

How does sample size relate to the amount of variance explained per SNP (significant “hit”)?

Genetic architecture of complex/multifactorial traits?

Genetic architecture describes the particular kinds of variants - rare vs common, and large effect vs small effect - that lead to a complex/multifactorial trait

What range of effect sizes do common variants identified in a GWAS typically have on the phenotype?

What about rare variants?

Common variants on GWAS: tend to have small effects

Rare variants on GWAS: tend to have large effects (ex. Mendelian diseases)

What do geneticists typically predict will be the relationship between effect size and allele frequency of detected hits?

What do we predict about common variants of large effect?

What about rare variants of small effects?

(Terms of GWAS)

High effect size = low frequency in the population (rare)

Low effect size = high frequency in the population (common)

Common variants of large effects will be easily identified in small sample sizes

Rare variants of small effect will require infinitely large sample sizes to identify

Common disease common variant model (CDCV)

Common diseases are the result of the cumulative (additive) effect of many variants that are common in the population.

Each person with the disease has a "high genetic burden" of these risk variants

Common disease rare variant model (CDRV)

Common diseases are the result of variants that are rare in the population.

Each person with the disease has a different rare variant of large effect.

On which model (CDCV or CDRV) is GWAS generally premised?

Does one model or the other more clearly reflect the reality of human multifactorial traits?

Do common variants tend to affect gene expression or gene product?

What about rare variants?

Common variants = gene expression

Rare variants = gene product

Missing heritability

Three possibilities that could account for missing heritability:

Which is the most likely option?

1. heritability estimates from twin studies: because twin studies are inflated

2. non-additive genetic effects

3. GWAS samples cannot detect small SNPs because they aren't large enough (most common)

In GWAS, how do we typically quantify the proportion of genetic variance accounted for in a phenotype?

How is a polygenic risk score computed?

What information is captured in a polygenic score?

We use a polygenic score to quantify the proportion of genetic variance accounted for in a phenotype

A polygenic risk score is a weighted linear sum of the SNPs in a large GWAS sample accounting for disease risk

It usually accounts for the strongest association SNPs, failing to account for any SNPs that might not have been statistically significant

SNP heritability

Measures the proportion of phenotypic variance explained by all SNPs

Total amount of SNPs accounted for in an analysis

Between increasing GWAS sample sizes, SNP heritability estimates, and inclusion of rare variants, is it possible to account for most of the so-called missing heritability?

No. We should be able to account for some of this missing heritability, not likely not all, by increasing sample sizes and including rare variants.

Pleiotropy

One genetic cause has many different effects on phenotype

How pleiotropy relates to the concept of a genetic correlation

There is significant genetic overlap between the genes correlated with psychological disorders (genetic correlation) like schizophrenia and major depressive disorder, or anorexia nervosa and OCD.

This suggests that these disorders may be more similar than initially thought, and may explain comorbidity of these disorders (e.g., people with anorexia nervosa tend to also have anxiety disorders)

What have geneticists found about the genetic correlations among multifactorial psychiatric phenotypes like schizophrenia, major depression, and bipolar disorder?

What does this suggest about diagnostic boundaries and underlying causal factors?

The causal (biological) mechanism for individual SNPs detected in a GWAS can be extremely difficult to ascertain. Why?

How can we nevertheless use GWAS results to infer causal mechanisms in terms of gene sets and pathways?

They have small effect sizes, which makes them difficult to ascertain

It is also hard to determine which is causal vs. correlated with causal mechanisms

How could GWAS results be clinically useful in the future?

- Calculating polygenic risk scores

- Providing potential therapeutic targets for pharmacological interventions

Population stratification:

Methods that can be used to address this problem:

Population stratification is an artifact, it occurs when different allele frequencies amongst populations become associated with the phenotypes of that population even when it is not a causal relationship

This has been done by focusing studies on just one ethnic group with shared ancestry.

Ancestry

Geographical origins of an individual's ancestors

Why do geneticists prefer to use ancestry rather than race or ethnicity?

Geographical origins of an individual's ancestors

Ethnicity does not have a biological basis and humans do not fit the biological definition of race (groups defined by independent morphology)

Conversely, ancestry recognizes different degrees of genetic relatedness among humans with those genetic differences being continuous rather than discrete

What do geneticists know about the distribution of common genetic variants among human ancestry populations?

What about rare or private variants?

Common variants tend to be ancient, shared across populations but potentially at different frequencies.

Common genetic variants exist in all (continental) human populations

Rare variants tend to be private and recent

Genetic drift defintion

Two types of genetic drift:

Do these phenomena tend to increase or decrease genetic variation in the resulting population?

1. Founder effects:

subset of population moving and ‘founding’ new population

2. Bottleneck: deadly event decreasing a population ex. Ashkanazi Jew genocide

These phenomena tend to decrease genetic variation

Why is it important for human geneticists to improve the genetic diversity of our samples?

Polygenic risk scores can become biased.

Results from the group that was overrepresented, when applied to an underrepresented group, can lead to assumptions of behaviors and comparisons of basic measures (like height) that don't match to reality

Restricting the diversity may not capture full range of genetic variation that is relevant to the trait being studied

How has restricting GWAS primarily to European populations in the past limited us in the present?