Late-onset Alzheimer?s disease (LOAD) affects a large portion of the human population and is highly heritable, though due to the difficulty of acquiring well-phentoyped data, genome-wide association studies (GWASs) of LOAD have had limited success in identifying associated genes. Additional statistical power would likely produce many discoveries related to the biology of LOAD, as it has for other complex phenotypes. This research plan proposes alternate data sources and new methods to increase the statistical power in genetic studies of LOAD. First, because LOAD is diagnosed late in life, large, cross-sectional studies cannot easily classify individuals as cases or controls. This limitation can be somewhat attenuated using pedigree information, as is done in the existing method, GWAX. Dr. Turley will extend GWAX to account for case-status, age, and other characteristics of both parents. These results will be meta-analyzed with available case-control- based results using Multi-Trait Analysis of GWAS (MTAG), leading to substantial gains in power and reduced risk of bias due to misclassification of cases. Second, LOAD and educational attainment (EA) have a genetic correlation of -0.3, suggesting that they may be associated with both common and unique biological pathways. Dr. Turley will seek to better understand LOAD by classifying and analyzing SNPs that are either jointly or uniquely associated with LOAD using Bayes-MTAG, an extension of MTAG that he is developing. Third, a lack of non-European GWAS cohorts have resulted in polygenic scores that perform poorly in those populations. Dr. Turley will develop Multi-Ancestry Meta-Analysis (MAMA), a trans-ethnic meta-analysis extension of MTAG that accounts for differences in linkage disequilibrium and genetic architecture across ancestries, to improve prediction of LOAD in non-European populations. The methods developed in each of these aims will increase statistical power, identifying novel loci, elucidating biological pathways, and improving polygenic prediction. Under the guidance his mentor, Dr. Benjamin Neale, his co-mentor, Dr. Xihong Lin, and a team of other advisers, Dr. Turley will pursue a rigorous program of training to accomplish the aims of this proposal and to develop into an independent researcher. The domains of this training include (i) epidemiology and genetics of aging, (ii) statistical and population genetics, (iii) large-scale data analysis and tools, and (iv) professional development. Development in these domains will be accomplished through coursework, attendance at conferences and workshops, experience leading teams and mentoring others, and regular feedback from his committee. Most importantly, the plan includes a detailed timeline, but which Dr. Turley and his mentoring team can monitor and evaluate progress. Overall, the training environment for the candidate is excellent, the mentors and advisors are world-class, the proposed studies address a crucial and timely unmet need, and the additional skills developed during this award will undoubtedly provide a strong foundation for the candidate to establish independent leadership in Alzheimer?s disease and statistical genetics.
Despite its high heritability, genome-wide studies of late-onset Alzheimer?s disease have had limited success in finding associated loci due to difficulties in recruiting sufficiently large samples. This application proposes using multiple sources of information to increase the statistical power including pedigree information, information for genetically-correlated phenotypes, and ancestry information. Results from this study will make novel discoveries about the biology of Alzheimer?s, elucidate causal mechanisms, and improve polygenic prediction.