Principal Investigator: Assistant Professor Mert Sabuncu
Massachusetts General Hospital, Radiology, 14729 Cambridge Street, Ap. 3, Cambridge MA 02139, United StatesTags: 13905, 32568, Genetic Correlation, heritability, Software, Statistical Analysis
Executive Committee on Research grant for $100,000 per yer over 5 years, only to be used for research at MGH
1a: We will use a novel software tool we have recently developed to compute the extent of the influence of human DNA on observable physical, clinical, and cognitive characteristics/traits (phenotype). Many of these traits are caused by genetic (heritable), environmental and life-style factors. Our primary aim is to identify those traits, where genetic factors play a significant role. This will enable us and other scientists to prioritize phenotypes for follow-up genetics studies. Our second aim will be to study the genetic overlap between phenotypes.
1b: Identifying the genetic factors that influence health-related, observable individual-level traits, such as disease diagnosis, will be critical for understanding the causal mechanisms of various clinical conditions, and developing prevention and treatment strategies. With rich phenotypic datasets such as the UK Biobank, it is going to be critical to prioritize phenotypes based on heritability. Those phenotypes which are largely determined by genetics (i.e., have large heritability) will be good candidates for further examining the underlying genetic causes.
1c: We will use a novel analytic strategy, which we recently published, to examine genome-wide marker (single nucleotide polymorphism, or SNP) data and phenotype data to examine the relationship between DNA and observable traits.
1d: We will use the full cohort.
“We will use novel statistical tools that we have recently developed to compute the heritability (proportion of phenotypic variation attributable to genetic variation in the population) for a broad spectrum of physical, clinical and cognitive traits, as well as regional and fine-grained neuroimaging measurements derived from structural and functional brain MRI scans. Identifying traits where genetic factors play a significant role will enable us and other scientists to prioritize phenotypes for follow-up molecular and statistical genetic analyses.
For complex traits that are found to be significantly heritable, we will further explore their genetic bases, which would involve conducting genetic correlation analyses across the phenome, genome-wide association analyses, polygenic risk analyses, and follow-up bioinformatic and functional analyses. For complex diseases that have a substantial genetic basis, such as heart attack, we will predict their occurrence by combining genetic data and risk factors using machine learning techniques (e.g., random forests and deep learning).”