Extended Data Fig. 4: Region-specific transcriptional profiles in the human subcortex.
From: TransBrain: a computational framework for translating brain-wide phenotypes between humans and mice
(a) The autocorrelation matrix of region-specific transcriptional embeddings in the human subcortex was divided into four sections. (Amy: Amygdala, red box; Hippo: Hippocampus, green box; BG: Basal ganglia, blue box; Tha: Thalamus, purple box). (b) Comparison of the hierarchical value of region-specific transcriptional embeddings. The hierarchy value, defined as the ratio of the mean similarity of each ROI to other ROIs within the same hierarchy versus the mean similarity to ROIs outside that hierarchy in the autocorrelation matrix (detached model vs shared model; Hippo (n = 4 ROIs): t(3) = 11.69, P = 0.0013; BG (n = 8 ROIs): t(7) = 2.61, P = 0.035; Tha (n = 8 ROIs): t(7) = 2.96, P = 0.021; two-sided paired-samples t-test). Box plots show the median (centre line), 25th and 75th percentiles (box bounds), and whiskers extending to 1.5 × IQR. (c) Comparison of the specificity of region-specific transcriptional embeddings. Similarity vectors for each ROI were normalized to a range of [0,1], sorted in descending order, and average (n = 22 ROIs). By comparing the differences in scaled similarity at the same rank, we assessed the specificity of transcriptional embeddings (detached model vs shared model: t(21) = -8.81, P = 1.70×10−8; two-sided paired-samples t-test). Lines show the mean. *P < 0.05, **P < 0.01, ***P < 0.001.
