Computationally modeling large-scale neural dynamics in autism using existing neuroimaging and transcriptomic datasets
- Awarded: 2019
- Award Type: Pilot
- Award #: 614955
Noninvasive neuroimaging, such as functional magnetic resonance imaging (fMRI), has provided insights into dysfunction of large-scale neural systems in autism spectrum disorder (ASD) and its relationship to cognitive deficits, yet it is unclear how such imaging biomarkers relate to cellular and synaptic features of ASD. A burgeoning multidisciplinary approach to bridge such gaps, known as ’computational psychiatry,’ leverages advances in theoretical neuroscience to investigate the links between mechanistic disturbances and emergent brain functions.
John Murray and Alan Anticevic will apply three innovative computational approaches to existing neuroimaging data sets in ASD to provide insight into the relationship between neural alterations and imaging biomarkers in ASD. Specifically, they plan to: (Aim 1) assess cortical circuit alterations in a mechanistic computational model of ASD, (Aim 2) examine brain-behavior relationships across individuals with ASD and (Aim 3) identify ASD-linked genes whose expression patterns align with neuroimaging-derived brain maps from individuals with ASD.
Aim 1 will apply biophysical models of neural circuit function to fit three cellular/circuit level mechanisms implicated in ASD: excitatory-inhibitory (E-I) imbalance, changes in the strength of local vs. long-range cortical interactions, and alterations in sensory-association cortical hierarchy. To do so, Murry and Anticevic will quantitatively fit the biophysical parameters of a large-scale neural circuit model using noninvasive human neuroimaging data (in particular, resting-state fMRI), leveraging publicly available ASD data sets. For cross-diagnostic insight, they will also apply this framework to data from the psychosis spectrum.
Aim 2 will characterize the neuro-behavioral geometry between ASD behavioral symptoms and neural variation. There is increasing evidence that ASDs have multidimensional relationships between symptoms and neural features. The goal here is to link neural and behavioral features in a framework that derives latent variables that link profiles of features across neural and symptom spaces.
Aim 3 will link human neuroimaging measures to brain-wide gene expression profiles, using the Allen Human Brain Atlas. Murray and Anticevic will examine how spatial gradients in transcriptional expression relate to ASD-relevant neuro-behavioral maps. This will allow them to link molecular targets to neuroimaging biomarkers in ASD. This in turn will directly facilitate future interrogations of circuit pathology and the design of novel therapeutics.