Despite great interest in using brain imaging to discover the genetic influences on the human brain, current gene discovery based on traditionally defined imaging phenotypes, such as total brain or hippocampal volume, are limited. The aim of this proposal is to use state-of-art dimension reduction methods to extract genetically orthogonal brain imaging phenotypes that will improve the power for gene discovery. The results can shed light on both understanding the genetic influences on normal development of human brain and, potentially, how the brain related disorders emerge from their genetic risks. We will use the UK biobank data to supplement our own imaging-genetics data resource (n ~= 1400 healthy individuals of European descent) to improve the power of our study to levels acceptable for genome-wide analysis. We will divide the combined cohorts into discovery, validation, and testing samples. We will use our models to define novel imaging phenotypes in the discovery sample that is intended to be the most genetically relevant. We will check the robustness of derived phenotypes in the validation sample and estimate the unbiased performance of final model in the independent testing samples. We are requesting imaging genomic data from UK Biobank which is approximately 1500 participants at present but we wish to receive later tranches of imaging data with corresponding genotyping information as it becomes available
