|
Abstract:
|
Neuroimaging has the potential to better illuminate the complex genetic basis of neuropsychological disorders, which have a biological basis rooted in brain function. Because they are quantitative, neuroimaging biomarkers avoid biases arising from imprecise clinical diagnostic criteria. To tackle high dimensionality and complex spatial relationships present in neuroimaging data, we used distance correlation tests in a genomewide association study predicated on multivariate diffusion tensor imaging measurements collected by the Pediatric Imaging, Neurocognition, and Genetics study. After correcting for multiplicity, distance correlation tests of the multivariate phenotype yield significantly greater power at detecting genetic variants affecting brain structure than mass univariate analysis of individual neuroimaging biomarkers. Furthermore, the distance correlation testing framework implicitly accounts for the covariance between neuroimaging phenotypes and is robust against stringent model specifications. These results demonstrate the potential of our approach to integrating and analyzing imaging and genomic data.
|
