Tractor: A framework allowing for improved inclusion of admixed individuals in large-scale association studies

Abstract

38Admixed populations are routinely excluded from medical genomic studies due to concerns over 39 population structure. Here, we present a statistical framework and software package, Tractor, to facilitate the 40 inclusion of admixed individuals in association studies by leveraging local ancestry. We test Tractor with 41 simulations and empirical data focused on admixed African-European individuals. Tractor generates ancestry-42 specific effect size estimates, can boost GWAS power, and improves the resolution of association signals. 43Using a local ancestry aware regression model, we replicate known hits for blood lipids in admixed 44 populations, discover novel hits missed by standard GWAS procedures, and localize signals closer to putative 45 causal variants. 46 47 48 Introduction 49Admixed groups, whose genomes contain more than one ancestral population such as African 50American and Hispanic/Latino individuals, make up more than a third of the US populace, and the population 51 is becoming increasingly mixed over time 1 . Many common, heritable, diseases including prostate cancer 2-5 , 52 asthma 6-9 , and several cardiovascular disorders such as atherosclerosis 10,11 are enriched in admixed 53 populations of the US. However, only a minute proportion of association studies address the genetic 54 architecture of complex traits in such groups 12,13 ; admixed individuals are systematically removed from many 55 studies due to the lack of methods and pipelines to effectively account for their ancestry such that population 56 substructure can infiltrate analyses and bias results [14][15][16][17][18][19][20][21] . Large-scale efforts to collect genetic data alongside 57 medically-relevant phenotypes are beginning to focus more on non-Eurasian ethnic groups that contain higher 58 amounts of admixture 22-27 , motivating the timely development of scalable methods to allow well-calibrated 59 statistical genomic work on these populations. If not addressed, this inability to analyze admixed people will 60 limit the clinical utility of large-scale data-collection efforts for minorities, exacerbating the concerning health 61 disparities that already exist 28-32 . 62In GWAS, the specific concern regarding including admixed participants is obtaining false positive hits due 63 to alleles being at different frequencies across populations. Most studies currently attempt to control for this by 64 using Principle Components (PCs) in a linear or linear mixed model framework. However, PCs capture broader 65 admixture fractions, and individuals' local ancestry makeup may differ between case and control cohorts even 66 if their global fractions are identical. Even including PCs as covariates, then, still leaves open the possibility for 67 false positive associations, as well as absorbing power. 68Studying diverse populations in gene discovery efforts not only reduces disparities but also benefits 69 genetic analysis for individuals of all ancestries. Perhaps the most notable example of this is in multi-ethnic 70 fine-mapping, which ca...

Improving polygenic prediction in ancestrally diverse populations

“…Although we have shown that this approach improved the prediction in both African and admixed American populations, indicating that PRS-CSx is robust to imperfectly matched LD reference panels, future work is needed to better model summary statistics from recently admixed populations. One promising direction is to characterize and incorporate local ancestries in PRS construction using ancestry-specific effect size estimates 33,34 . The Bayesian modeling framework of PRS-CSx and the flexibility of continuous shrinkage priors also allow for the incorporation of functional annotations and fine-mapping results into PRS construction to improve the portability of PRS, as shown by recent studies 35 .…”

Incorporating European GWAS findings improve polygenic risk prediction accuracy of breast cancer among East Asians

“…Since the genomes of admixed individuals are a mosaic of segments with different ancestral origins, a first step would be to get ancestry‐specific effect size estimates and P ‐values from training GWASs, which is often not available from publicly available summary statistics. If individual‐level training GWAS data is available, recently developed methods like Tractor (Atkinson et al, 2020) could be applied to obtain ancestry‐specific summary statistics by generating ancestry dosage at each site from local ancestry inference calls and running a local ancestry‐aware regression. Similarly, for the validation data, local ancestry haplotype dosage for each person at each variant need to be estimated and weighted by the ancestry‐specific effect size estimated in the previous step to allow the generation of “ancestry‐specific” PRSs.…”

Continuity of Genetic Risk for Aggressive Behavior Across the Life-Course

“…The overrepresentation of EA participants in GWAS is partly because admixed populations were long considered inconvenient in gene discovery studies, as this led to population stratification issues. However, due to advances in GWAS methods, populations with mixed genetic backgrounds can now be included in GWAS to obtain accurate estimates of SNP effects, boost power, and improve fine-mapping of effects by leveraging linkage disequilibrium differences (Asimit et al 2016;Atkinson et al 2020). Beside the general benefits to gene discovery studies, the inclusion of diverse genetic backgrounds will improve our understanding of genetic liability across diverse populations, as demonstrated for example in a study of glycemic traits (Chen et al 2021), where 30% of the participants were of non-European ancestry.…”