Principal Investigator: Dr Mert Sabuncu
Department: Massachusetts General HospitalTags: 13905, featured
Using the UK Biobank data, we have completed an analysis which estimates the heritability for a wide spectrum of phenotypes under our approved data request (ref: 13905) (Ge et al. bioRxiv 2016). We now plan to extend our initial analysis to estimate the heritability of phenotypes derived from multi-modal brain MRI scans. In addition to the derived measurements that the UK Biobank has computed and distributed (e.g. sizes of individual brain structures), we also plan to explore other measurements computable from the volumetric image data (e.g., cortical thickness, surface area, functional connectivity profiles) and estimate high-resolution maps for the heritability of brain structure and function, as we demonstrated in our prior work (Ge et al. PNAS 2015). We thus request access to full brain MRI images (NIFTI files). Also, based on the feedback we received on our initial phenome-wide heritability analysis, we would like to add all blood assay results (category 100081) to our analysis.
Our objective is at least two-fold. First, we will employ heritability as a metric to compare several standard processing software that yield measurements derived from brain MRI scans. For example, we would like to quantify the heritability of volumetric measurements computed with FreeSurfer, a popular MRI processing pipeline, and FIRST (FMRIB’s Integrated Registration and Segmentation Tool). Similarly there are several widely used processing pipelines for diffusion and functional MRIs. The UK Biobank currently provides a select few set of measurements for each imaging modality. We hope to examine the heritability values of a range of measurements that we can compute using various pipelines. Second, we plan to generate voxel-level and surface-based maps of heritability. For instance, we would like to estimate the heritability of cortical thickness and folding. We are interested in conducting heritability analyses for various demographic groups (e.g. young vs old), and seeking interactions with various environmental factors as well. We also plan to conduct these types of analyses for examining the genetic overlap between image-derived measurements and cognitive, clinical and behavioral phenotypes.