Genetic and Environmental Variance Components of White Matter Tissue Properties Reveal Genetic and Environmental Factors in Individual Variability
McKenzie Hagen, University of Washington
Using non-invasive neuroimaging tools, we can collect measurements of white matter tissue properties, which facilitate long-range communication between brain regions. In order to answer psychological questions about behavior, we can use these brain tissue property measurements to explain individual differences in human behavior and cognition. At the same time, we want to be able to describe the biological mechanisms that form this relation, and also estimate our confidence in any relationships that we draw between brain measurements, biology and behavior.
For example, individual differences in white matter tissue properties explain some of the individual differences in key behavioral traits, such as cognitive abilities. One of the biological mechanisms that drives these individual differences is the genetic heritability of brain tissue properties, which can be estimated in samples that contain individuals that are genetically related. Heritability of behavioral traits can also be estimated in these samples.
Analyses of heritability of brain tissue properties typically rely on separate univariate estimates of heritability for each individual brain tissue property measurement. However, brain tissue property measurements are high-dimensional multivariate data that contain large degrees of interdependence and covariance. To accommodate this data structure, we developed analysis methods that leverage the multidimensional structure in the data to decompose the total covariance in white matter properties into genetic and environmental influences. We then used supervised learning methods to assess the predictive ability of the separated genetic and environmental contributions to white matter properties covariance in predicting a range of different behavioral phenotypes. We evaluated the reliability of our models with bootstrap resampling. With these methods, we hope to bridge the gap between the genetic influence on white matter properties and the predictive ability of white matter to demonstrate biological mechanisms that form the relation between white matter tissue properties and behavioral phenotypes.
Abstract Author(s): McKenzie Hagen, Keshav Motwani, Jason Yeatman, Eardi Lila, Ali Shojaie, Ariel Rokem