Back

Distinct transcriptomic and hierarchical organization associated with brain hyperconnectivity and hypoconnectivity in autism spectrum disorder

Vairam, A.; Bhavna, K.; Uddin, L. Q.; Misic, B.; Roy, D.

2026-08-01 neuroscience
10.64898/2026.07.31.742105 bioRxiv
Show abstract

Functional hyperconnectivity and hypoconnectivity in autism spectrum disorder (ASD) are typically treated as opposing expressions of a single circuit-level disturbance, but their molecular and hierarchical basis remains unclear. We combined resting-state fMRI from 1,737 individuals from the Autism Brain Imaging Data Exchange (ABIDE I/II) with gene-expression maps from the Allen Human Brain Atlas and show that hyperconnectivity and hypoconnectivity are dissociable neurobiological phenomena, differing in molecular signatures, cortical-hierarchical embedding, age-group profile, and cognitive associations, rather than a single connectivity axis. Hyperconnectivity was concentrated in higher-order cortical and cerebellar regions and was greater in older participants, while hypoconnectivity was consistent across age groups and localized to subcortical and orbitofrontal systems. ASD was associated with reorganization of the sensory-to-transmodal cortical gradient, most pronounced in association networks. Hyperconnectivity- and hypoconnectivity-associated genes showed partially distinct neurotransmitter profiles and differential embedding within cortical hierarchy, both enriched in transmodal cortex and linked to social-cognitive, perceptual, attentional, and reward-related functions. This dissociation was preserved across developmental stage, sex, and symptom severity. These findings indicate hyperconnectivity and hypoconnectivity are not two poles of one process but two separable components of a reproducible molecular-hierarchical architecture, offering a multi-scale framework linking transcriptomic organization to systems-level brain dysfunction in ASD.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.