Advancing Genetic Studies in Latin American Populations: Enhancing Imputation and Investigating X Chromosome Associations with Alzheimer’s Disease

Abstract

Estimating genetic risk factors of diseases is challenging in diverse populations underrepresented in genomic studies. Genome-wide association studies (GWAS) are commonly used to discover genetic risk loci associated with health and disease. However, the majority of participants in GWAS are of European descent. These studies rely on imputation and many existing imputation reference panels are largely composed of individuals of European ancestry, resulting in lower imputation quality in underrepresented populations. The studies comprising this dissertation investigate how the composition of imputation reference panels affects imputation quality in four target Latin American cohorts. We achieve this through comparing imputation quality for chromosomes 7 and X when altering the imputation reference panel by: 1) increasing the number of Latin American individuals; 2) excluding either Latin American, African, or European individuals, or 3) increasing the Indigenous American (IA) admixture proportions of included Latin Americans. We found that increasing the number of Latin Americans in the reference panel improved imputation quality in the four cohorts; however, for some there were differences between chromosomes 7 and X. Finally, increasing Indigenous American (IA)-like admixture proportions in the reference panel increased imputation quality at different levels in the populations. The difference in imputation results between populations and chromosomes suggests that existing and future reference panels containing Latin American individuals are likely to perform differently in different Latin American populations, consistent with what we know of the varying population structure and ancestry proportions of the Latin Americans. As a further investigation, we imputed 87,393 variants on the X chromosome for 49,405 Latin American individuals and conducted a GWAS and X chromosome-wide association study (XWAS) in a Caribbean cohort. We identified 3 autosomal and 17 X-chromosome variants significantly associated with Alzheimer’s disease. Future studies are required to replicate X chromosome findings and include more robust X chromosome data for more diverse Latin American populations. This work highlights the importance of not treating Latin Americans as a homogenous population and taking population structure and the X chromosome into account when studying these populations.