Neuroplasticity in adolescence is a period of critical development for self-processing and emotional regulation, disruptions to which are linked to persistent depression. A recent task-based, neurofeedback fMRI study of depressed vs. healthy adolescents elicited differential functional connectivity (FC) amongst brain regions of interest (ROI). However, results were methodologically unadjusted for inherent inter-subject variability in task-based fMRI. To address this, a random covariance model (RCM) for bi-level subject and group-specific graphical models was applied to analyze FC in 17 pre-identified ROIs for (1) depressed vs. healthy, (2) suicide attempters vs. other, and (3) across the entire AAL3 atlas. RCMs were tuned with modified BIC and significance assessed with FDR-corrected 5000-permutation tests. Results in (1) corroborated significant hyper- and hypo-activation of R. and L. Amygdala-Cuneus connectivity, respectively, in depressed vs. healthy adolescents. Initial results in (2) also suggest significant Amygdala-Cuneus activation in suicide attempters vs. other, while (1) suggests exploratory results in (3) ought to yield novel, whole-brain FC differences between groups.