Multivariate Analyses in Behavioural Genetics PDF

Multivariate Analyses in Behavioural Genetics 04 December 2023 14:59 Main Ideas Multivariate genetic designs Multivariate twin model Co-occurrence Co-morbidity Heterogeneity Notes Some questions can be addressed using a multivariate genetic design Notes Extending the simple ACE model to estimate the...

Multivariate Analyses in Behavioural Genetics 04 December 2023 14:59 Main Ideas Multivariate genetic designs Multivariate twin model Co-occurrence Co-morbidity Heterogeneity Notes Some questions can be addressed using a multivariate genetic design Notes Extending the simple ACE model to estimate the heritability of latent factors Notes Genetic correlations with academic achievement ○ Questions about the possible common aetiology of different traits ▪ For example: □ What extent is the covariation between different traits influenced by genetic and environmental factors □ Do different traits share common genetic and environmental influences □ What extent do genetic and environmental factors contribute to longitudinal stability and change in a single trait □ To what extent do genetic and environmental factors contribute to the longitudinal association between multiple traits Multivariate twin model Development ○ Can be extended by looking at shared latent factors ○ MZ and DZ correlations - Univariate version ▪ Univariate twin model: Inspecting the cross-twin within-trait covariance □ Higher MZ than DZ correlations indicate genetic influence □ DZ correlations more than half MZ indicate shared environmental influence □ MZ correlations less than 1 indicate nonshared environmental influence □ Multivariate findi ○ Genetic sou ▪ “Gene ▪ Appar ○ Example: Depress ○ Question to ○ Twin correla Comorbidity Comorbidity is the norm - in the Dunedin study, over 80% of those who experienced one psychiatric disorder experienced at least one other psychiatric disorder (Caspi et al., 2020) Example: Anxiety & Depression ▪ Example: ADHD fr ○ Question : w ○ Study on Hy □ □ Extending the simple ACE model to study the longitudinal development of one trait: The Cholesky decomposition ○ ○ Study on ina ○ MZ and DZ correlations – Multivariate scenario ▪ Multivariate twin model: (additionally) comparing the magnitude of MZ and DZ correlations across traits and/or across time points Multivariate beha ○ Risk and pro ○ Different int ○ More resear Studying comorbidity with latent factor models: Phenotypic results Extending the simple ACE model to study the longitudinal development of multiple traits: The cross-lag panel model □ ○ Heterogeneity ○ □ Can compare correlation of different traits within the same twin □ within-twin cross-trait correlations - common influences? □ Cross-twin cross-trait correlations - common influences = familial? ○ Path analysis of twin data (the ACE model) ▪ Univariate twin models: Used to estimate the sources of variance in a trait ▪ Multivariate twin models: Used to estimate the sources of covariance between traits or time points (“bivariate” if 2) ○ Cross lag effect can indicate a causal effect after accounting for the covariation Studying comorbidity with latent factor models: Twin results Important findings from multivariate genetic research concerning co-occurrence, comorbidity, development and heterogeneity ○ Example - Academic achievement in different subjects (Rimfield et al., 2015) (Univariate results) Example: ASD ○ Diagnostic c Social □ □ □ Co-morbidity: Molecular genetic results ▪ ○ ○ Genetic and environmental correlations ▪ Genetic correlations = Overlap in underlying genetic factors Pleiotropy (genes influence multiple traits) ▪ Environmental correlations = Overlap in underlying environmental factors Range between 0 (independent influences) and 1 (complete overlap) ▪ Can be positive (influences act in the same direction) or negative (influences act in the opposite direction) Phenotypic correlations with academic achievement Non-s □ □ □ ○ Study: Twin ○ □ Summary Multivariate genetic designs allow us to look beyond whether something is heritable or not For example, we can study aetiology of : ○ Co-morbidity ○ Developmental continuity and change ○ Origins of different subtypes Implications: ○ Basic science (e.g. gene hunting) ○ Applied psychology (treatment research and practice) PSYC0036 Genes and Behaviour Page 1 Notes ngs on co-occurrence and comorbidity: Implications urces of co-occurrence and comorbidity imply the presence of “generalist genes”: etic diagnoses” can differ from symptom -based diagnoses. rently distinct conditions might benefit from similar intervention approaches (prevention/treatment). ○ Multivariate results (teacher ratings, female participants) ▪ sion from childhood to adolescent address with multivariate design: what drives developmental change and comorbidity ations rom childhood to adolescence what drives longitudinal stability yperactivity/impulsivity: (Pingault et al., 2015): Baseline level (intercept): 90% of the variance was explained by additive genetic influences. Linear systematic change (slope): 81% of the variance was explained by additive genetic influences, of which 37% shared with the intercept. attention avioural genetic analyses of development: Implications otective factors are relevant to specific developmental stages terventions may be required at different developmental stages rch on developmental trajectories in families with multiple vulnerabilities ○ Example: Heterogenity ▪ Conduct problems Low levels of callous-unemotional traits (LCU): ® Often aggress when feel under threat ® Feel bad about hurting others ® Can have high levels of anxiety High levels of callous-unemotional traits (HCU): ® View proactive aggression as rewarding ® Do not worry about hurting others ® Have low levels of anxiety ▪ Conduct problems and high CU traits : Affective processing □ Atypical processing of other people’s distress (fear and sadness), possibly also happiness and disgust □ Report feeling little fear □ Are less reactive to punishment in standard learning tasks and in intervention settings ▪ Conduct problems and low CU traits : Affective processing □ Hostile attribution bias □ Oversensitive to perceived anger (even with neutral stimuli) □ ▪ Results: Antisocial behaviour and CU traits criteria: domain: ‘Has unusual eye gaze, facial expression or gestures’ ‘Has at least one good friend’ (reversed item) ‘Has odd style of communication; old-fashioned, formal, or pedantic’ social domain: ‘Is extremely distressed by changes to routine or familiar arrangements’ ‘Has a strong interest in an unusual topic’ ‘Notices small details others might miss’ n correlations (teacher ratings) PSYC0036 Genes and Behaviour Page 2 □ Notes Multivariate behavioural genetic analyses of heterogeneity: Implications ○ Different subtypes of a given disorder may have different aetiology ○ Partly distinct genetic and environmental risk factors for children with the same disorder ○ Intervention tailored to genotypic and neurocognitive risk? Treatment and policy implications