This article summarizes the findings and recent corrections to the UCLA ATLAS Community Health Initiative study, originally published in Genome Medicine. The initiative represents a significant step toward diversifying genomic medicine by linking electronic health records (EHR) with genetic data from over 36,000 individuals in the Los Angeles area.
Addressing the Diversity Gap in Genomics
Historically, genomic research has been heavily biased toward populations of European descent. This "diversity gap" limits the applicability of precision medicine to underrepresented groups, potentially exacerbating health disparities. The UCLA ATLAS initiative seeks to bridge this gap by utilizing the immense diversity of the Los Angeles population to study how genetic ancestry and self-identified race/ethnicity (SIRE) influence disease risk.
Methodological Framework
To analyze this complex dataset, researchers integrated EHR data with genomic information using several key techniques:
- Principal Component Analysis (PCA): Used to visualize genetic relationships and cluster individuals alongside the global 1000 Genomes Project reference panel.
- ADMIXTURE Analysis: Estimated the proportions of ancestral populations within individual genomes.
- Logistic Regression: Calculated Odds Ratios (OR) for various health conditions, adjusting for self-identified categories to isolate the predictive power of genetic ancestry.
- GWAS & PheWAS: Identified specific genetic variants associated with diverse clinical phenotypes.
Mapping Global Diversity in a Local Biobank
A central achievement of the ATLAS initiative is its ability to capture a broad spectrum of global genetic diversity within a single metropolitan health system.
As shown in Figure 2, the ATLAS cohort successfully captures the full spectrum of global genetic diversity, with participants filling the "genetic space" between major continental reference populations. The analysis reveals that while SIRE and Genetically Inferred Ancestry (GIA) often correlate, they capture distinct information. GIA provides a more granular understanding of the genetic "mosaic" in admixed populations, such as Hispanic/Latino and African American groups. Interestingly, the study also found that cultural markers like "preferred language" can serve as effective proxies for genetic sub-clusters, particularly among East Asian American participants.
Ancestry-Specific Disease Risk
The integration of genomic data with EHRs allowed researchers to quantify how disease prevalence varies across ancestry groups. Key findings include:
- Skin Cancer: Strongly associated with European Ancestry (EA), with an Odds Ratio (OR) of approximately 6.0.
- Uterine Leiomyoma: Showed a significantly higher risk in African American (AA) groups (OR > 2.0).
- Chronic Kidney Disease: Elevated risk was observed in both African American and Hispanic/Latino (HL) groups (OR ~2.0).
These results underscore the importance of moving beyond broad racial categories to identify specific genetic drivers of health disparities.
Limitations and the Path to Precision Medicine
Despite its scale, the study faced limitations, such as small sample sizes for rare conditions in non-European groups and potential biases inherent in EHR data (e.g., healthcare access affecting the number of clinical encounters). Future work will focus on identifying functional genetic variants that drive these observed disparities, moving the field closer to truly equitable precision medicine.
Summary updated