Behavior Genetics The Study of Variation and Heredity Prof. Timothy Bates tim.bates@ed.ac.uk

Overview Introduce behavioural genetic methodology Adoption studies Twin studies Family studies Introduce concepts of Genetic variance Genetic correlation Examples of cognitive ability

Motivation: Why look at genetics?

Causality: genetics uniquely untangles causality Does Parental Neglect cause Anti-Social Behavior? Eaves et al., 2010

Why aren’t we all the same? σ2 = variance Sum((x-mean)2) > 0

Applies to Continuous Variation…

and to categorical variation: Thresholded Continuous trait

In BG, categories like “depression” are often viewed as thresholds on a liability scale

Why do things run in families? Shared genes + Shared environment

Genes, Environment and Intelligence 07/03/2014 Genes, Environment and Intelligence Sir Francis Galton (Victorian polymath) Half cousin of Charles Darwin Genealogical studies of eminent families Hereditary Genius (1869) Intelligence is inherited "Twins have a special claim upon our attention; it is, that their history affords means of distinguishing between the effects of tendencies received at birth, and those that were imposed by the special circumstances of their after lives."

Behavioural Genetics Methods 07/03/2014 Behavioural Genetics Methods Use known genetic relatedness to estimate variability in a trait due to genes (heritability) Children share 50% of genes with each parent Identical twins share 100% Non-identical twins/siblings 50% Grandparents 25%, cousins 12.5% Unrelated people, no genetic sharing

Resemblance for Cognitive Ability in family members 22/11/2016 Resemblance for Cognitive Ability in family members Behavioral Genetics (2016). Valerie S. Knopik,‎ Jenae M. Neiderhiser,‎ John C. DeFries,‎ Robert Plomin

Estimating h2 from extended Pedigree Studies 22/11/2016 Estimating h2 from extended Pedigree Studies Components of genetic and environmental variance are estimated by univariate linear mixed models Luciano et al., (2010) Intelligence, 38: 304-313.

Estimating h2 from Adoption Studies 07/03/2014 Estimating h2 from Adoption Studies If adopted offspring are more similar to biological than adoptive parent indicates genetic influence e.g., adoptees with schizophrenia show concordance rate of 12% with biological family members versus 3% with adoptive family members Example to the right is height MZ twin- pair members in Tom Bouchard’s MISTRA study of twins reared apart

Similarity for cognitive ability across adoptive family classes 22/11/2016 Similarity for cognitive ability across adoptive family classes

Estimating h2 from Twin Studies 07/03/2014 Estimating h2 from Twin Studies Compare trait similarity in identical and fraternal twins 1/250 births are identical twins 1/150 are fraternal twins

Twins Reared Apart Design MZs reared apart σ2 = A+C+E cov= A (=h2) Modern example Tom Bouchard Minnesota Study of Twins Reared Apart (MISTRA)

Genes, Environment and IQ 07/03/2014 Genes, Environment and IQ

The Classical Twin Design: Variance decomposition MZ: Monozygotic Share 100% of genes (A) Share family (C) Have unique experiences (E) DZ: Dizygotic Share (on average) 50% of genes (1/2 A) σ2 = A+ C + E rMZ = A + C rDZ = .5A + C E = 1 – rMZ A = 2 * (rMZ – rDZ) C = 1 – (A + E) Falconer’s formula

22/11/2016 Heritability The proportion (ratio) of population variance that can be attributed to genetic influences This genetic contribution usually represents the cumulative effects of many genes (“polygenic” “complex trait genetics”) Narrow sense: Broad sense:

Genetic Variation Genetic additivity (A): Allelic effects sum within and across loci Genetic non-additivity 1: (Dominance): Interaction of the effects of alleles at a locus Not shared between parents and offspring Genetic non-additivity 2: (Epistasis) Interaction of the effects of alleles across loci

Twin Modelling in SEM (Structural Equation Modelling) ACE model

Specifying the ACE model Use two groups: MZs (with A1 <-> A2 path = 1.0) DZs, with A1 <-> A2 path = .5

SEM for Twin Modelling OpenMx umx R package, on CRAN Open Source: http://openmx.ssri.psu.edu umx Open Source High-level assistance for twin and general modeling umxACE()

Specifying the ACE model in OpenMx

How heritable is height? library(umx) & umxACE() data(twinData); df = twinData selDVs = c("ht1", "ht2") mzData <- df[df$zygosity %in% "MZFF", ] dzData <- df[df$zygosity %in% "DZFF", ] m1 = umxACE(selDVs= selDVs, dzData= dzData, mzData= mzData)

ACE Model Assumptions Twins are ‘ordinary folks’ 22/11/2016 ACE Model Assumptions Twins are ‘ordinary folks’ Similar to singletons despite greater gestational difficulties No particular effects of growing up side by side No assortative mating MZ and DZ twins experience environmental influences equally

22/11/2016 ACE Model Assumptions Genetic and environmental influences are independent and additive No gene-environment correlation Exposure to environmental conditions depends on an individual's genotype No gene-environment interaction Environmental exposure does not moderate effects of genes on behaviour

Genes, Environment and IQ 07/03/2014 Genes, Environment and IQ Gene by environment interaction Heritability value can change depending on the group(s) measured e.g. Higher heritability in high SES vs low SES (near or below poverty line in the US) Turkheimer E et al. Psychological Science (2003) 623- 628 Bates T C et al. Psychological Science (2013) 2111- 2116

Genes, Environment and IQ 07/03/2014 Genes, Environment and IQ Heritability generally increases with the age of the sample measured: ~0.30 in childhood to 0.50 in middle adulthood to 0.70 in old age Haworth et al. (2010) Mol Psychiatry, 15: 1112–1120.

More than one Trait Traits covary We can explain association via: ADHD E1 E2 A1 A2 a11 IQ rG C1 C2 rC 1 rE c11 a22 c22 e11 e22 Traits covary We can explain association via: Additive genetic factors (rG) Shared environment (rC) Non-shared environment (rE) e.g. Kuntsi et al. (2004) Am J Med Genet B, 124:41

Cholesky Decomposition Twin 1 Phenotype 1 A1 A2 E1 E2 a11 a22 e11 e22 1 Phenotype 2 C1 C2 c11 c22

Cholesky Decomposition Twin 1 Phenotype 1 A1 A2 E1 E2 a11 a21 a22 e11 e21 e22 1 Phenotype 2 C1 C2 c11 c21 c22

Cholesky Decomposition Twin 1 Phenotype 1 A1 A2 E1 E2 a11 a21 a22 e11 e21 e22 1 Phenotype 2 C1 C2 c11 c21 c22 Twin 2 1/0.5

Genetic Correlation

rG vs. robserved If the rG = 1, the two sets of genes overlap completely If however a11 and a22 are near to zero, genes do not contribute (much) to the observed correlation The contribution to the observed correlation (phenotypic) is a function of rG and both heritabilities

Proportion of rP due to additive genetics: ADHD and IQ a2 x IQ rg 1 a2y Heritability of ADHD Genetic correlation between ADHD and IQ Heritability of IQ Phenotypic corr. ADHD & IQ Proportion of pheno.corr. due to genetic factors

Multivariate Modelling Luciano et al., (2001). Behavior Genetics, 31(6), 581-592.

(Many) Other important Types of Models Common Pathway Model Independent Pathways Model Longitudinal Data Latent Growth models Latent Factor models Interaction models G*E Measured Environment, Measured Genes Direction of Causation

Latent Growth Model AI CI EI As 2 Cs Es G1 G2 Genotypes G3 Phenotype Time 1 AI a11 c11 e11 Intercept Time 2 Time 3 AS2 CS2 ES2 AS3 CS3 ES3 AS1 CS1 ES1 CI EI As a22 c22 e22 Slope 2 Cs Es a21 c21 e21 im sm as11 as22 as33 cs11 es11 cs22 es22 cs33 es33 Twin 1 Bi Bs Phenotype Age G1 G2 G3 Genotypes

Quantitative and Qualitative Sex Effects Females aF = aM? cF = cM? eF = eM? Does the magnitude of genetic/environmental effects differ in males and females? (QUANTITATIVE) With OS pairs: Are the same genes/environments acting in males and females (QUALITATIVE) Compare correlations from DZ like-sex DZ opposite-sex pairs Sex-specific effects suggested when OS DZr < Same-sex DZr.

Bivariate Direction of Causation Modelling

Extending theTwin Design Adding other relatives Siblings Cousins Children of twins Parents, grandparents Uncles, Aunts…

Thank you! Questions Please!