Network level analysis methods for the developing connectome

Determining the mechanisms by which the human brain generates cognition, perception, and emotion hinges upon quantifying the relationships between coordinated brain activity and behavior. Contemporary connectome research views the brain as a large-scale, complex network composed of nonadjacent, yet connected brain regions. These connectomes are frequently composed of tens of thousands of connections between brain regions which poses a challenge for identifying biologically meaningful and reproducible associations with behavioral or clinical outcomes. In this presentation I will cover approaches to connectome-wide association analysis that leverage the inherent systems-level architecture of the human brain in order to probe fundamental biological mechanisms. Specifically, our lab has implemented methods from the genome-wide association literature, frequently referred to as ‘pathway analysis,’ ‘enrichment analysis,’ or ‘overrepresentation analysis’ in a comprehensive and flexible software toolbox for use by the neuroscience community. However, while a large body of research has demonstrated reproducible systems-level architecture in data from awake children and adults, research suggests that these models are ill posed for use with data from babies under the age of three which are most commonly collected during natural sleep. I will present recent work benchmarking system-level model fit across early brain development and provide recommendations for assessing connectome-wide associations early in development.