Expression quantitative trait loci (eQTL) mapping, colocalization and transcriptome-wide association studies (TWAS) are widely used to contextualize genome-wide association study loci, yet the impact of upstream RNA-sequencing processing choices on these analyses remains critically unexplored. Two poorly characterized decisions include the transcriptomic reference annotation used for quantification of short-read RNA-sequencing samples, which defines transcript composition and structure, and the quantification method itself. Here, through a comprehensive analysis spanning 48 tissues in GTEx and 7 widely studied and diverse phenotypes, we find that these choices substantially affect eQTL detection and gene expression prediction with significant downstream impacts on colocalization and TWAS results. Concordant results across pipelines are further enriched for longer genes, genes with more isoforms, and greater expression heritability. Our findings demonstrate that seemingly minor methodological decisions substantially affect these common analyses, highlighting the need for standardized practices to ensure reproducible genetic association studies.

Wild emmer wheat (Triticum turgidum ssp. dicoccoides, BBAA), the wild progenitor of cultivated tetraploid and hexaploid wheats, provides an important reference for investigating centromere reorganization during polyploidization and domestication. Here we generated a near-complete genome assembly of wild emmer wheat, including centromeric regions for all 14 centromeres. Comparative analyses with cultivated tetraploid and hexaploid wheats revealed extensive variation in centromere-associated sequences and centromere-proximal structure, as well as distinct subgenome-associated patterns in the A and B subgenomes. The A subgenome was characterized by positional shifts and sequence turnover of CRW-rich centromere-derived blocks, whereas the B subgenome showed reduced CentT566 satellite abundance accompanied by localized retention or expansion on specific chromosomes. These results provide a high-resolution tetraploid wheat reference and reveal subgenome-associated patterns of centromere remodeling during wheat polyploid evolution.

Transcriptome-wide association studies (TWAS) are widely used to identify genes involved in complex traits and to infer the direction of gene effects on traits. However, despite their popularity, it remains unclear how accurately TWAS recover the true direction of a gene's effect on a trait. Here we estimate the false sign rate of TWAS for plasma proteins, leveraging the expectation that increased gene expression should generally increase protein expression. We then extend this framework to complex traits, where loss-of-function burden tests provide the expected direction of effect. In both analyses, we observe high discordance with expectations, with TWAS showing a false sign rate of 23% for plasma proteins and 33% for complex traits. We propose that much of the sign discordance arises when expression quantitative trait loci in non-trait-relevant tissues tag haplotypes associated with traits in genome-wide association studies via distinct, tightly linked regulatory variants, yielding spurious TWAS associations with the correct genes but with unreliable direction of effect. These findings show that TWAS-based direction-of-effect estimates should be interpreted with caution and raise concerns about the reliability of TWAS more broadly.

Traditional genome-wide association studies have primarily focused on detecting main genotype effects, often overlooking genotype-environment interactions (GxE), which are essential for understanding context-specific genetic effects and refining disease etiology. Here we present fast genotype-environment interaction analysis (fastGxE), a scalable and effective genome-wide GxE method designed to identify genetic variants that interact with environmental factors to influence traits of interest. fastGxE controls for polygenic effects, polygenic interaction effects and noise heterogeneity; it remains robust to the number of environmental factors involved in GxE interactions; and it ensures scalability for genome-wide GxE analysis in large biobank studies, achieving speed improvements of 32.98-126.49 times over existing approaches. We illustrate the benefits of fastGxE through extensive simulations and an in-depth analysis of 32 physical traits and 67 blood biomarkers from the UK Biobank. In real-data applications, fastGxE identifies 7 genomic loci associated with physical traits, including 4 novel ones, and 18 genomic loci associated with blood biomarkers, 12 of which are novel. The new discoveries highlight the dynamic interplay between genetics and the environment, uncovering potentially clinically relevant pathways that could inform personalized interventions and treatment strategies.

Characterizing noncoding variant function remains an important challenge in human genetics. Genomic deep learning models have emerged as a promising approach to enable in silico prediction of variant effects. These models include supervised sequence-to-activity models, which predict molecular phenotypes such as genome-wide chromatin states or gene expression levels directly from DNA sequence, and self-supervised genomic language models. Here we review progress in leveraging these models for noncoding variant effect prediction. We describe practical considerations for making such predictions and categorize the types of ground-truth data used to evaluate variant effect predictions, providing insight into the settings in which current models are most useful. Our Review highlights key considerations for practitioners and opportunities for improvement in model development and evaluation.

We performed a large-scale genome-wide association study meta-analysis of magnetic resonance imaging-derived liver triglyceride content across 71,488 European-ancestry participants from the UK Biobank, approximately doubling the sample size of previous genome-wide association studies[1]. We identified 60 independent loci, including 31 not previously reported, and estimated the single nucleotide polymorphism-based heritability of liver triglyceride content at 25%. Integrative fine-mapping and gene-prioritization analyses nominated candidate effector genes for 48 loci. Of the 31 newly identified variants, 14 were validated in meta-analyses of clinically diagnosed metabolic dysfunction-associated steatotic liver disease (42,163 cases and 1,646,254 controls). People in the highest versus lowest decile of a polygenic risk score for liver triglyceride content had 2.71-fold higher odds of cirrhosis and 7.47-fold higher odds of hepatocellular carcinoma. Plasma proteins associated with the liver triglyceride polygenic risk score were enriched for metabolic remodeling and inflammatory pathways, with liver-enriched protein signatures increasing and visceral adipose signatures decreasing with higher genetic risk. Together, these findings refine the genetic architecture of liver triglyceride content, linking polygenic susceptibility to downstream metabolic and inflammatory pathways and providing insights into the genetic basis of steatotic liver disease.

Patterns of gene and sequence preservation across species provide insight into gene essentiality under natural selection. Here we developed two interspecies constraint metrics across 462 mammalian species: Gene Identity Score of Mammalian Orthologs (GISMO), which quantifies gene loss weighted by evolutionary distance to humans, and GISMO-mis, which measures missense constraint using the ratio of missense to synonymous variation across mammals. Both metrics are strongly correlated with human loss of function, missense and dosage constraint derived from large-scale sequencing studies, yet are largely independent of coding sequence length. GISMO and GISMO-mis identify essential genes missed by existing human-only metrics, particularly short genes. Genes in the most constrained deciles are enriched for rare variant associations in disorders linked to reduced fecundity, including schizophrenia, autism and neurodevelopmental disorders, and show strong enrichment for common variant heritability. These deciles also exhibit reduced copy number variation, consistent with dosage sensitivity. In addition, GISMO improves prioritization of recessive disease genes and captures signatures of homozygous selection across the mammalian tree. Taken together, GISMO and GISMO-mis leverage deep mammalian evolution to provide complementary, length-robust measures of gene essentiality that strengthen disease-gene prioritization and illuminate the selective forces shaping human biology.

The genetic architecture of inborn errors of immunity continues to grow in complexity, exposing the limitations of variant-centric concepts that have long shaped the field. While complete loss-of-function genotypes have been instrumental in defining essential gene functions in human immunity and inflammation, they have also entrenched binary interpretations of causality that fail to account for pleiotropy, variable expressivity and incomplete penetrance-especially in monoallelic disease. Monoallelic variants often perturb signaling rather than abolishing it. The integration of these perturbations into molecular and cellular systems, and eventually immune circuits, governs phenotypic expression. Here, we revisit and extend classic concepts of genetic dominance and genetic pleiotropy and suggest that these are not sole properties of alleles but emerge from the complex and dynamic behavior of molecular, cellular and immune systems to buffer genetic variance.

Conditional and joint (COJO) analysis of genome-wide association study (GWAS) data is a standard post-GWAS analysis for identifying statistically independent associations. Although GWAS meta-analyses are increasingly conducted across multiple ancestry groups, the optimal strategy for multiancestry COJO analysis remains unclear. Here we introduce Manc-COJO, a method for multiancestry COJO analysis. Simulations and real-data analyses show that Manc-COJO improves detection of independent association signals while reducing false positives compared with single-ancestry COJO implemented in genome-wide complex trait analysis (GCTA) and its ad-hoc multiancestry adaptations. We also introduce Manc-COJO:FIX, a follow-up algorithm to identify ancestry-specific independent associations. The C++ implementation of Manc-COJO is over 120-fold faster than GCTA-COJO and supports linkage disequilibrium references derived from either individual-level genotype data or pre-computed matrices, facilitating analyses when data sharing is limited.

H2AK119ub (H2Aub) and H3K27me3, which are deposited by the Polycomb group (PcG) complexes PRC1 and PRC2, respectively, cooperatively repress developmental genes. However, in human preimplantation embryos, H3K27me3 undergoes genome-wide erasure and subsequent restoration. How H3K27me3 is reestablished and whether PcG complexes still regulate gene repression in human embryos remain unknown. Here, we profiled H2Aub dynamics from human gametes to blastocysts. H2Aub emerges in GC-rich regions as early as the two-cell stage, progressively accumulates at both active and inactive genes, and further culminates at PcG targets in later stages. H3K27me3 is weakly correlated with H2Aub in oocytes and two-cell embryos but colocalizes with H2Aub at H3K4me3-depleted, inactive PcG targets in blastocysts. Loss of H2Aub abolishes H3K27me3 reestablishment at PcG targets causing moderate derepression, accompanied by ectopic H3K27me3 deposition and impaired embryonic development. Finally, H2Aub and DNA methylation cooperatively maintain H3K27me3 patterning in human embryonic stem cells. Therefore, H2Aub-dependent H3K27me3 deposition underlies the earliest establishment of Polycomb repression in humans.