Towards a fine-scale population health monitoring system

Abstract

Understanding population health disparities is an essential component of equitable precision health efforts. Epidemiology research often relies on definitions of race and ethnicity, but these population labels may not adequately capture disease burdens specific to sub-populations. Here we propose a framework for repurposing data from Electronic Health Records (EHRs) in concert with genomic data to explore enrichment of disease within sub-populations. Using data from a diverse biobank in New York City, we genetically identified 17 sub-populations, and noted the presence of genetic founder effects in 7. By then linking community membership to the EHR, we were able to identify over 600 health outcomes that were statistically enriched within a specific population, with many representing known associations, and many others being novel. This work reinforces the utility of linking genomic data to EHRs, and provides a framework towards fine-scale monitoring of population health.

Exome Sequencing Reveals a High Prevalence ofBRCA1andBRCA2Founder Variants in a Diverse Population-Based Biobank

“…(1:66) and lowest in those of Hispanic/Latino descent (1:283). We previously used genotype array data to identify fine-scale population groups in BioMe using genetic ancestry (23), revealing eight communities with greater than 400 individuals represented ( Table 2). Across these, prevalence was highest in individuals with AJ ancestry (1:49), among whom the majority (72 out of 80 individuals, or 90.0%) harbored one of the three AJ founder variants (c.5266dupC and c.68_69delAG in BRCA1, and c.5946delT in BRCA2), and 8 individuals (10.0%) harbored a different variant in BRCA1/2 ( Supplementary Table S3).…”

“…Sample preparation and exome sequencing were performed at the Regeneron Genetics Center as previously described (22) yielding N=31,250 samples and n=8,761,478 sites. Genotype array data using the Illumina Global Screening Array was also generated for each individual (23). Post-hoc filtering of the sequence data included filtering of N=329 low-quality samples, including low coverage, contaminated and genotype-exome discordant samples; N=208 gender discordant and duplicate samples were also removed.…”

“…Self-reported ancestry categories were derived from a multiple-choice survey administered to participants upon enrollment into the BioMe Biobank (23). Participants could select one or more of the following categories: African-American/African, American Indian/Native American, Caucasian/White, East/Southeast Asian, Hispanic/Latino, Jewish, Mediterranean, South Asian/Indian, or Other.…”

“…These Finally, we determined the proportion African genetic ancestry in mixed ancestry Hispanic/Latino populations using the ADMIXTURE (24) software. We assumed five ancestral populations (k=5) with 5-fold cross validation across n=256,052 SNPs in N=27,984 unrelated participants that were also genotyped on the Global Screening Array (GSA), in addition to N=4,149 reference samples representing 5 continental regions (23). We estimated relatedness using the software KING (25), and for all prevalence estimates in self-reported and genetic ancestry groups, we excluded second degree relatives and above.…”

A review of UMAP in population genetics

“…Figure 4a and b visualize, respectively, the Genome Aggregation Database (gnomAD v3) from the Broad Institute [10] and Biobank Japan (BBJ) [11,12], each of which contains over 100, 000 individuals. When applied to ethnically diverse groups such as the UKB, BioMe [13], and the Million Veterans Program (MVP) [14], UMAP tends to highlight groups with different international migration and admixture histories. In relatively more homogeneous populations such as BBJ, it highlights clusters related to geographic features such as island populations.…”