Multivariate genetic analyses of developmental complex traits

Student thesis: Doctoral ThesisDoctor of Philosophy


Consistent evidence from the quantitative genetic literature points towards a substantial genetic component underlying variation and covariation across complex traits. Novel genomic methods coupled with large population-based samples, afford the possibility to leverage this information to tackle important developmental questions. This thesis focuses on multivariate genetic and genomic approaches as applied to polygenic prediction and inference of trait associations across development, with a focus on cognitive- and psychopathology-related traits. The thesis follows two main themes:
Polygenic prediction
Methods that leverage the covariance structure of genetically correlated traits to increase power for variant discovery can in turn be used to boost predictive power of polygenic scores. In a first study I compare several multi-trait genomic approaches in the context of polygenic prediction of cognitive-related traits (general intelligence and educational achievement) in childhood and adolescence (Chapter 2). As genetic predictors become more powerful, they can be employed to further our understanding of the gene-environment interplay underlying variation in common complex traits. In a second study (Chapter 3), I focus on the longitudinal prediction of educational achievement by constructing penalized multivariable prediction models integrating multiple polygenic scores and environmental predictors, gaining insights into their multivariate interplay (gene-environment correlation and interaction).
Developmental co-occurrence of psychopathology
Akin to the concept of general intelligence, the co-occurrence of traits related to mental health during development suggests a general dimension of psychopathology underlying the emergence and co-morbidity of problem behaviours in childhood (the p-factor). In a third study (Chapter 4) I systematically investigate the manifestation of the p-factor across childhood and adolescence by means of multivariate genetic methods, showing that this co-occurrence is partly explained by a common genetic aetiology. There are at least two plausible processes that can account for the co-occurrence of psychopathology traits in childhood. First, as investigated in Chapter 4 the correlation between psychopathologies could be the product of individual differences between people on stable traits attributable to a heritable p-factor. Second, the developmental co-occurence of psychopathology could emerge from a causal process within people where the temporal state on one variable causally influences the state of another variable, inducing correlation between them. To this end, in a fourth study I investigate longitudinal reciprocal effects between problem behaviours (Chapter 5), separating between vs within person effects in two longitudinal population-based cohorts. Extending this model to family-level data, I further investigate reciprocal directional influences between siblings over time, separating them from similarities between siblings that arise through shared (genetic or environmental) influences that exist in a family. The thesis concludes (Chapter 6) with a discussion of future prospects for multivariate genomic research, opportunities for integrating emerging methods, challenges and limitations. 
Date of Award1 Mar 2021
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorRobert Plomin (Supervisor) & Jean-Baptiste Pingault (Supervisor)

Cite this