GWAS

Privacy-preserving framework for multi-site GWAS on quantitative traits using a distributed linear mixed model and randomized encoding so servers never see raw genotypes or phenotypes—only obfuscated intermediates—while improving speed versus several cryptographic baselines.

Privacy-preserving GWAS, multi-site, quantitative traits, federated analysis

Genome-wide association studies help uncover genetic influences on complex traits and diseases. Importantly, multi-site data collaborations enhance the statistical power of these studies but pose challenges due to the sensitivity of genomic data. Existing privacy-preserving approaches to performing multi-site genome-wide association studies rely on computationally expensive cryptographic techniques, which limit applicability. To address this, we present PP-GWAS, a privacy-preserving algorithm that improves efficiency and scalability while maintaining data privacy. Our method leverages randomized encoding within a distributed framework to perform stacked ridge regression on a linear mixed model, enabling robust analysis of quantitative phenotypes. We show experimentally using real-world and synthetic data that our approach achieves twice the computational speed of comparable methods while reducing resource consumption.

regenie is a C++ program for whole genome regression modelling of large genome-wide association studies. It is developed and supported by a team of scientists at the Regeneron Genetics Center.

whole genome regression

Genome-wide association analysis of cohorts with thousands of phenotypes is computationally expensive, particularly when accounting for sample relatedness or population structure. Here we present a novel machine-learning method called REGENIE for fitting a whole-genome regression model for quantitative and binary phenotypes that is substantially faster than alternatives in multi-trait analyses while maintaining statistical efficiency. The method naturally accommodates parallel analysis of multiple phenotypes and requires only local segments of the genotype matrix to be loaded in memory, in contrast to existing alternatives, which must load genome-wide matrices into memory. This results in substantial savings in compute time and memory usage. We introduce a fast, approximate Firth logistic regression test for unbalanced case-control phenotypes. The method is ideally suited to take advantage of distributed computing frameworks. We demonstrate the accuracy and computational benefits of this approach using the UK Biobank dataset with up to 407,746 individuals.

R framework that links GWAS signals to single-cell-defined cell types via a cell-type gene specificity score (expression magnitude and consistency) and regression on gene-level association statistics, with influential-gene follow-up for interpretability.

GWAS, scRNA-seq, cell type, MAGMA, post-GWAS interpretation

Integrating single-cell RNA sequencing with Genome-Wide Association Studies (GWAS) can uncover cell types involved in complex traits and disease. However, current methods often lack scalability, interpretability, and robustness. We present seismic, a framework that computes a novel specificity score capturing both expression magnitude and consistency across cell types and introduces influential gene analysis, an approach to identify genes driving each cell type-trait association. Across over 1000 cell-type characterizations at different granularities and 28 polygenic traits, seismic corroborates known associations and uncovers trait-relevant cell groups not apparent through other methodologies. In Parkinson's and Alzheimer's, seismic unveils both cell- and brain-region-specific differences in pathology. Analyzing a pathology-based Alzheimer's GWAS with seismic enables the identification of vulnerable neuron populations and molecular pathways implicated in their neurodegeneration. In general, seismic is a computationally efficient, powerful, and interpretable approach for mapping the relationships between polygenic traits and cell-type-specific expression, offering new insights into disease mechanisms.

Recruitment of ~500,000 UK adults with questionnaire and physical measures; genome-wide array genotyping on the full cohort underpins most early GWAS and PRS applications.

Joint host whole-genome sequencing and gut microbiome profiling in 252 Chinese individuals with graded cognitive disability. Microbiome GWAS for latent enterosignature (Anaerostipes-enriched) abundance; integrates AD polygenic risk and brain cell-type expression context. Open-access report in Microbiome.

Genome-wide association analysis in 8,956 German individuals links host variation to single-taxon and overall gut microbiome composition; replicates ABO histo-blood group and FUT2 secretor associations with Bacteroides and Faecalibacterium. Includes Mendelian randomization for IBD-relevant microbial effects.

Meta-analysis of genome-wide associations between host genotypes and gut microbial structural variants (deletion dSVs and variable vSVs) from metagenomes in four Dutch cohorts (n = 9,015), with replication in Tanzania (n = 279). After frequency filters, GWAS used 3,552 common SV phenotypes aggregated across 49 bacterial species with sufficient metagenomic coverage. Highlights ABO/FUT2–GalNAc pathways and Faecalibacterium prausnitzii SVs.

Perspective on microbial GWAS (mbGWAS): state of the art, heterogeneity of microbiome assays, power, and directions for genetic analysis of the gut microbiome.

Japanese shotgun metagenome and SNP-array GWAS (7,213,469 post-imputation variants, MAF >1%, Minimac4 Rsq >0.7): microbial traits included species, gene orthologs, and pathways. Two study waves—dataset 1 (gut microbiome, plasma metabolome, genotype): n = 300; dataset 2 (gut microbiome, genotype): n = 224—were combined by fixed-effect meta-analysis (total n = 524; Figure S1 / Table S1). Species GWAS used 423 microbial species (study-wide threshold 5×10⁻⁸/423); headline hits include PDE1C–Bacteroides intestinalis and TGIF2 / TGIF2-RAB5IF–B. acidifaciens, plus microbial gene ortholog associations with blood group A conditioned on East Asian FUT2 secretor status. Metabolome arm in dataset 1 supports microbiome–plasma integration. Public data resource.

MiBioGen consortium meta-analysis: genome-wide host genotypes with 16S fecal microbiome data across 24 cohorts (n = 18,340). N_MICROBES = 410 genus-level groups in the MiBioGen framework (paper: nine genera detected in >95% of samples). Identifies 31 genome-wide significant loci affecting microbial taxa (e.g. lactase LCT, ABO, fucosyltransferase cluster).

16S rRNA amplicon-based GWAS of salivary microbiota traits in unrelated Danish adults from the ADDITION-PRO cohort (n = 610). Identifies host SNPs associated with oral bacterial abundance and beta diversity; several variants link to metabolic traits. Oral (not gut) microbiota — listed here for host–microbiome genetics.

Harmonized shotgun metagenome GWAS in 16,017 adults from four Swedish studies, with replication in 12,652 participants from the Norwegian HUNT study. Reports loci including OR51E1–OR51E2 (microbial richness), LCT, ABO, FUT2, MUC12, CORO7–HMOX2, SLC5A11, FOXP1, FUT3–FUT6, and species-level associations.