COVID-19 has caused numerous infections with diverse clinical symptoms. To identify human genetic variants contributing to the clinical development of COVID-19, we genotyped 1457 (598/859 with severe/mild symptoms) and sequenced 1141 (severe/mild: 474/667) patients of Chinese ancestry. We further incorporated 1401 genotyped and 948 sequenced ancestry-matched population controls, and tested genome-wide association on 1072 severe cases versus 3875 mild or population controls, followed by trans-ethnic meta-analysis with summary statistics of 3199 hospitalized cases and 897,488 population controls from the COVID-19 Host Genetics Initiative. We identified three significant signals outside the well-established 3p21.31 locus: an intronic variant in FOXP4-AS1 (rs1853837, odds ratio OR = 1.28, P = 2.51 × 10⁻¹⁰, allele frequencies in Chinese/European AF = 0.345/0.105), a frameshift insertion in ABO (rs8176719, OR = 1.19, P = 8.98 × 10⁻⁹, AF = 0.422/0.395) and a Chinese-specific intronic variant in MEF2B (rs74490654, OR = 8.73, P = 1.22 × 10⁻⁸, AF = 0.004/0). These findings highlight an important role of the adaptive immunity and the ABO blood-group system in protection from developing severe COVID-19.

Estimation of relatedness between pairs of individuals is important in many genetic research areas. When estimating relatedness, it is important to account for admixture if this is present. However, the methods that can account for admixture are all based on genotype data as input, which is a problem for low-depth next-generation sequencing (NGS) data from which genotypes are called with high uncertainty. Here, we present a software tool, NGSremix, for maximum likelihood estimation of relatedness between pairs of admixed individuals from low-depth NGS data, which takes the uncertainty of the genotypes into account via genotype likelihoods. Using both simulated and real NGS data for admixed individuals with an average depth of 4x or below we show that our method works well and clearly outperforms all the commonly used state-of-the-art relatedness estimation methods PLINK, KING, relateAdmix, and ngsRelate that all perform quite poorly. Hence, NGSremix is a useful new tool for estimating relatedness in admixed populations from low-depth NGS data. NGSremix is implemented in C/C++ in a multi-threaded software and is freely available on Github https://github.com/KHanghoj/NGSremix.

Standard measures of linkage disequilibrium (LD) are affected by admixture and population structure, such that loci that are not in LD within each ancestral population appear linked when considered jointly across the populations. The influence of population structure on LD can cause problems for downstream analysis methods, in particular those that rely on LD pruning or clumping. To address this issue, we propose a measure of LD that accommodates population structure using the top inferred principal components. We estimate LD from the correlation of genotype residuals and prove that this LD measure remains unaffected by population structure when analyzing multiple populations jointly, even with admixed individuals. Based on this adjusted measure of LD, we can perform LD pruning to remove the correlation between markers for downstream analysis. Traditional LD pruning is more likely to remove markers with high differences in allele frequencies between populations, which biases measures for genetic differentiation and removes markers that are not in LD in the ancestral populations. Using data from moderately differentiated human populations and highly differentiated giraffe populations we show that traditional LD pruning biases FST and principal component analysis (PCA), which can be alleviated with the adjusted LD measure. In addition, we show that the adjusted LD leads to better PCA when pruning and that LD clumping retains more sites with the retained sites having stronger associations.

Principal component analysis (PCA) is widely used in statistics, machine learning, and genomics for dimensionality reduction and uncovering low-dimensional latent structure. To address the challenges posed by ever-growing data size, fast and memory-efficient PCA methods have gained prominence. In this paper, we propose a novel randomized singular value decomposition (RSVD) algorithm implemented in PCAone, featuring a window-based optimization scheme that enables accelerated convergence while improving the accuracy. Additionally, PCAone incorporates out-of-core and multithreaded implementations for the existing Implicitly Restarted Arnoldi Method (IRAM) and RSVD. Through comprehensive evaluations using multiple large-scale real-world data sets in different fields, we show the advantage of PCAone over existing methods. The new algorithm achieves significantly faster computation time while maintaining accuracy comparable to the slower IRAM method. Notably, our analyses of UK Biobank, comprising around 0.5 million individuals and 6.1 million common single nucleotide polymorphisms, show that PCAone accurately computes the top 40 principal components within 9 h. This analysis effectively captures population structure, signals of selection, structural variants, and low recombination regions, utilizing <20 GB of memory and 20 CPU threads. Furthermore, when applied to single-cell RNA sequencing data featuring 1.3 million cells, PCAone, accurately capturing the top 40 principal components in 49 min. This performance represents a 10-fold improvement over state-of-the-art tools.

Greenlandic Inuit and other indigenous populations are underrepresented in genetic research^1,2, leading to inequity in healthcare opportunities. To address this, we performed analyses of sequenced or imputed genomes of 5,996 Greenlanders with extensive phenotypes. We quantified their historical population bottleneck and how it has shaped their genetic architecture to have fewer, but more common, variable sites. Consequently, we find twice as many high-impact genome-wide associations to metabolic traits in Greenland compared with Europe. We infer that the high-impact variants arose after the population split from Native Americans and thus are Arctic-specific, and show that some of them are common due to not only genetic drift but also selection. We also find that European-derived polygenic scores for metabolic traits are only half as accurate in Greenlanders as in Europeans, and that adding Arctic-specific variants improves the overall accuracy to the same level as in Europeans. Similarly, lack of representation in public genetic databases makes genetic clinical screening harder in Greenlandic Inuit, but inclusion of Greenlandic data remedies this by reducing the number of non-causal candidate variants by sixfold. Finally, we identify pronounced genetic fine structure that explains differences in prevalence of monogenic diseases in Greenland and, together with recent changes in mobility, leads to a predicted future reduction in risk for certain recessive diseases. These results illustrate how including data from Greenlanders can greatly reduce inequity in genomic-based healthcare.

The COVID-19 pandemic has caused over 220 million infections and 4.5 million deaths worldwide. Current risk factor cannot fully explain the diversity in disease severity. Here, we present a comprehensive analysis of a broad range of patients' laboratory and clinical assessments to investigate the genetic contributions to COVID-19 severity. By performing GWAS analysis, we discovered several concrete associations for laboratory traits and used Mendelian randomization (MR) analysis to further investigate the causality of traits on disease severity. Two causal traits, WBC counts and cholesterol levels, were identified based on MR study, and their functional genes are located at genes MHC complex and ApoE, respectively. Our gene-based analysis and GSEA revealed four interferon pathways, including type I interferon receptor binding and SARS coronavirus and innate immunity. We hope that our work will contribute to studying the genetic mechanisms of disease and serve as a useful reference for COVID-19 diagnosis and treatment.