Genomics and Biobank Research for Health Equity

National Institute On Minority Health And Health Disparities

Investigators

Linked publications, trials & patents

Abstract

Today we are starting to fulfill the promise of genomics for personalized medicine and healthcare. Interestingly, and for a variety of reasons, publicly funded databases continue to miss a vast portion of the world's genetic variation. As of January 2018, the GWAS catalog has registered 78% of individuals from European ancestry while underrepresented populations make up less than 4% including African (2.4%), Hispanic or Latin American (1.3%) and Native American (0.03%). The sampling bias is referred to as the genomics research gap and has the potential to exacerbate existing health disparities among underrepresented and underserved populations. Populations underrepresented in biomedical research bear a disproportionate burden for many diseases, including diabetes type 2. This is particularly true for Native American communities in the US, who have among the lowest levels of participation in genomic studies seen for any ethnic group. Genomic studies must be more representative of all populations so that all people can benefit from the upcoming genomic revolution in healthcare. We are interested in studying how combinations of genetic ancestry in admixed Latin American populations may impact genomic determinants of health and disease. We have found ancestry-enriched SNPs in Latin American populations having a substantial effect on health- and disease-related phenotypes. Current methods for genetic ancestry (GA) inference do not scale to biobank-size genomic datasets. We present Rye-a new algorithm for GA inference at biobank scale. We compared the accuracy and runtime performance of Rye to the widely used RFMix, ADMIXTURE and iAdmix programs and applied it to a dataset of 488221 genome-wide variant samples from the UK Biobank. Rye infers GA based on principal component analysis of genomic variant samples from ancestral reference populations and query individuals. The algorithm's accuracy is powered by Metropolis-Hastings optimization and its speed is provided by non-negative least squares regression. Rye produces highly accurate GA estimates for three-way admixed populations-African, European and Native American-compared to RFMix and ADMIXTURE, and shows 50 runtime improvement compared to ADMIXTURE on the UK Biobank dataset. We also developed the UK Biobank (UKB) Health Disparities Browser with the aims of (i) facilitating the exploration of the landscape of health disparities in the UK and (ii) directing the attention to areas of disparities research that might have the greatest public health impact. Health disparities were characterized for UKB participant groups defined by age, country of residence, ethnic group, sex and socioeconomic deprivation. We defined disease cohorts for UKB participants by mapping participant International Classification of Diseases, Tenth Revision (ICD-10) diagnosis codes to phenotype codes (phecodes). For each of the population attributes used to define population groups, disease percent prevalence values were computed for all groups from phecode case-control cohorts, and the magnitude of the disparities was calculated by both the difference and ratio of the range of disease prevalence values among groups to identify high- and low-prevalence disparities. We identified numerous diseases and health conditions with disparate prevalence values across population attributes, and we deployed an interactive web browser to visualize the results of our analysis: https://ukbatlas.health-disparities.org. The interactive browser includes overall and group-specific prevalence data for 1513 diseases based on a cohort of >500 000 participants from the UKB. Our group is also developing algorithms and software for local and global ancestry estimation that scale well with biobank data and are actively maintaining a repository on GitHub: https://github.com/healthdisparities.