Fig. 2: The leading population mode tracks decreased cognitive functioning in CNV carriers. | Nature Biomedical Engineering

Fig. 2: The leading population mode tracks decreased cognitive functioning in CNV carriers.

From: A pattern-learning algorithm associates copy number variations with brain structure and behavioural variables in an adolescent population cohort

Fig. 2

a, CNVs significantly impact the revealed dominant behaviour pattern. Cohen’s d values of canonical scores calculated between controls and, separately, carriers of deletions and duplications are plotted for the first canonical mode. Filled dots indicate significantly different average scores from controls, based on cross-validation testing. b, Brain region correlates reveal a whole-brain pattern. Brain loadings were calculated as the correlation between brain scores and 148 regional brain volumes. These loadings are mapped onto the cortical surface to illustrate their spatial distribution, with colours indicating the strength and direction of the associations (red, positive; blue, negative). The accompanying bar plot highlights the 20 regions with the strongest loadings, with 95% confidence intervals estimated by rerunning the PLS model on 1,000 bootstrap resamplings of participants (cf. Methods). G, gyrus; S, sulcus. The radial bar chart shows the average brain loadings in each of the seven large-scale networks defined by Schaefer–Yeo parcellation. c, Behaviour correlates highlight real-life functioning. Behaviour loadings were calculated as the correlation between behaviour scores and ~1,000 behaviour measures. Left: behaviour loadings from a PLS model, grouped by category and colour coded accordingly. Each dot represents an individual phenotype, with the y axis indicating its effect (loading) strength and direction. Right: summary of the results presented by the average absolute loading for each category in a circular bar plot, capturing the overall contribution of each domain to the derived behaviour score. In summary, the first canonical mode highlights the connection between frontoparietal and temporal regions and assessments of cognition and demographics.

Source data

Back to article page