Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript.
Deep learning shows promise for predicting gene expression levels from DNA sequences. However, recent studies show that current state-of-the-art models struggle to accurately characterize expression variation from personal genomes, limiting their usefulness in personalized medicine.
Variants in the HLA region on chromosome 6 are strongly associated with many immune-related diseases. A method to construct personalized HLA genomes from single-cell RNA sequencing data, coupled with single-cell HLA expression quantitative trait loci modeling, identifies how genetic variants influence HLA gene expression across cell states.
Single-cell transcriptomes and single-cell chromatin accessibility profiles generated using EasySci provide a global view of aging and Alzheimer’s pathogenesis-associated cell population dynamics in human and mouse brains.
Analysis of single-nucleus RNA sequencing and single-nucleus assay for transposase-accessible chromatin with sequencing data derived from synovium of patients with rheumatoid arthritis identifies regions with dynamic accessibility that correlate with cell states. Dynamic peaks are more strongly enriched for autoimmune disease heritability.
A downsampling approach to assess causal variant fine-mapping, replication failure rate, finds that commonly used methods may be miscalibrated. Simulations suggest this is probably due to a nonsparse genetic architecture model misspecification. Incorporating infinitesimal effects in the SuSiE and FINEMAP frameworks improves performance.
This Perspective article explores complex synthetic lethal relationships in cancer, which can involve several partners. Understanding this complexity presents challenges and opportunities for the development of therapeutics that target these interactions.
Cross-ancestry genome-wide association meta-analyses identify new risk loci for peptic ulcer diseases and provide evidence that gastrointestinal cell differentiation and hormone regulation contribute to their etiology.
Neural networks are a common machine learning architecture for predicting phenotype from genomic sequence. This analysis finds that they err in calling the variant direction of effect, with important implications for personalized predictions.
Population analysis of 516 wild and domesticated broomcorn millet genomes and a graph-based pangenome based on de novo assemblies of 32 representative accessions identify genomic variations associated with domestication traits.
Roulette enables the estimation of germline mutation rates at basepair resolution from humans. Genes encoding small nuclear RNA showed significant deviations from the mutation rate predicted by Roulette, highlighting RNA polymerase III (Pol III)-dependent transcription as a potent source of mutations in the human genome.
A test of four genomic sequence-to-expression deep learning models (Enformer, Basenji2, ExPecto, Xpresso) finds that they often fail to predict the correct direction of effect of cis-regulatory genetic variation on gene expression.
The chemotherapeutic agent CX-5461 is shown to be a potent mutagen in hTERT-RPE1, HAP1 and human induced pluripotent stem cells. The compound generates distinct mutational patterns of single- and double-base substitutions, as well as of small insertions and deletions, that were detectable following a single exposure.
scHLApers is an analysis pipeline that quantifies single-cell expression of HLA genes using a personalized genomic reference. Mapping of HLA expression quantitative trait loci at single-cell resolution identifies dynamic effects across cell states.
A multivariate framework for isoform-resolution transcriptome-wide association studies enables modeling of a greater number of genes, with the benefit of identifying isoform-specific associations with psychiatric traits not observed at the gene level.
Genome-wide analyses yield insights into the polygenic effects contributing to clinical heterogeneity in attention deficit hyperactivity disorder, advancing understanding of its genetic etiology and serving as a model for future studies in other complex disorders.
Whole-genome sequencing data of individuals from the UK Biobank and Iceland and a somatic mutation barcoding strategy enabled detection of clonal hematopoiesis at scale. This comprehensive study provides insights into the epidemiology, somatic and germline genetics, and disease associations of clonal hematopoiesis.
