Imputation Performance in Latin American Populations: Improving Rare Variants Representation With the Inclusion of Native American Genomes.

Current Genome-Wide Association Studies (GWAS) rely on genotype imputation to increase statistical power, improve fine-mapping of association signals, and facilitate meta-analyses. Due to the complex demographic history of Latin America and the lack of balanced representation of Native American genomes in current imputation panels, the discovery of locally relevant disease variants is likely to be missed, limiting the scope and impact of biomedical research in these populations. Therefore, the necessity of better diversity representation in genomic databases is a scientific imperative. Here, we expand the 1,000 Genomes reference panel (1KGP) with 134 Native American genomes (1KGP + NAT) to assess imputation performance in Latin American individuals of mixed ancestry. Our panel increased the number of SNPs above the GWAS quality threshold, thus improving statistical power for association studies in the region. It also increased imputation accuracy, particularly in low-frequency variants segregating in Native American ancestry tracts. The improvement is subtle but consistent across countries and proportional to the number of genomes added from local source populations. To project the potential improvement with a higher number of reference genomes, we performed simulations and found that at least 3,000 Native American genomes are needed to equal the imputation performance of variants in European ancestry tracts. This reflects the concerning imbalance of diversity in current references and highlights the contribution of our work to reducing it while complementing efforts to improve global equity in genomic research.

Genotype imputation via matrix completion.

Most current genotype imputation methods are model-based and computationally intensive, taking days to impute one chromosome pair on 1000 people. We describe an efficient genotype imputation method based on matrix completion. Our matrix completion method is implemented in MATLAB and tested on real data from HapMap 3, simulated pedigree data, and simulated low-coverage sequencing data derived from the 1000 Genomes Project. Compared with leading imputation programs, the matrix completion algorithm embodied in our program MENDEL-IMPUTE achieves comparable imputation accuracy while reducing run times significantly. Implementation in a lower-level language such as Fortran or C is apt to further improve computational efficiency.