Fig. 1: Linking genome-wide CNVs and brain–behaviour population covariation.
a, Genome-wide CNV identification in the ABCD population cohort. We investigated 7,338 children from the ABCD database. In total, 5,314 children do not carry any protein-coding CNV, 486 carry a deletion and 1,406 carry a duplication fully encompassing one or more genes. A total of 132 participants carried both deletion and duplication (left plot, outer circle). The ratio of males and females is similar in every group (left plot, inner circle). The histogram (right plot) depicts the age of participants across CNV groups. b, Overlaid histograms showing the distribution of the number of genes encompassed by each CNV. c, A partial least squares model links the brain with behaviour in one holistic model. We estimate a multivariate relationship structure among 148 brain atlas volume measures and ~1,000 behaviour measures spanning 20 categories based on measurements from children without any CNV. The canonical scores represent the latent variable expressions calculated from linear combinations of the original brain and linear combinations of behaviour measurements that maximize the covariance between the two sets of variables. The number in brackets represents the number of phenotypes per category. d, CNV status associated with individual expression strengths of brain and behaviour patterns. The bar plot shows the average brain and behaviour scores of CNV carriers across all modes. Error bars represent 95% confidence intervals based on 1,000 bootstrap resampling replications. Stars indicate significant differences identified through cross-validation testing from control participants who were not used to derive model parameters (cf. Methods). These results reveal that carrying a CNV significantly impacts canonical scores across different modes of brain–behaviour covariation, emphasizing the utility of a multivariate holistic framework that cuts across single disciplines.
