Reviews & Analysis

Liability scores for chronic obstructive pulmonary disease obtained from our deep learning model improve genetic association discovery and risk prediction. We trained our model using full spirograms and noisy medical record labels obtained from self-reporting and hospital diagnostic codes, and demonstrated that the machine-learning-based phenotyping approach can be generalized to diseases that lack expert-defined annotations.

We developed a machine learning model to quantify cardiac fibrosis (which is associated with cardiovascular disease) using cardiac MRI data from 41,505 UK Biobank participants. In the subsequent large-scale GWAS of cardiac fibrosis, we identified 11 independent genomic loci, 9 of which were implicated in in vitro cardiac fibroblast activation.

Reconstructing phylogenetic trees from large collections of genome sequences is a computationally challenging task. We developed MAPLE, a method for performing phylogenetic inference on large numbers of closely related genomes, which might be useful when studying the evolution and spread of SARS-CoV-2 and of infectious pathogens in future pandemics.

Genome assembly of nine wild species and two domesticated accessions of tomato generated a super-pangenome for the tomato clade. Comparative analyses revealed the landscape of structural variations in wild and cultivated tomatoes and led to the discovery of a wild tomato gene that has the potential for yield increase in modern breeding.

Specific chromatin features, especially histone H3 lysine 27 acetylation, are widely used to identify active enhancers, yet current methods are imprecise. New work suggests that histone H2B N terminus multisite lysine acetylation (H2BNTac) is a notable signature of active enhancers and could substantially improve enhancer prediction.

A study uses single-cell RNA sequencing to profile human papillomavirus (HPV)-positive and -negative oropharyngeal squamous cell carcinoma, revealing considerable diversity within and between tumors. Within HPV-positive tumors, subsets of malignant cells are found with undetectable HPV expression and decreased HPV-related phenotypes, which may influence prognosis and response to therapy.

A GWAS meta-analysis, combined with tracing the parental transmitted and non-transmitted alleles in parent–offspring pairs, enabled us to distinguish the effects of maternal and fetal genomes on gestational duration and their links with birth weight. The identified genes are more likely to be differentially expressed during labor and show signs of antagonistic pleiotropy with fetal genome effects on birth weight.

Transposable elements (TEs) are transcriptionally activated in cancer and can generate chimeric TE–gene transcripts that are tumor specific. Our pan-cancer study reveals that these transcripts are a source of tumor-specific antigens that are shared across tumor types, bind to HLA and are presented on the extracellular surface of cancer cells appended to membrane-bound proteins.

Telomere length is an important determinant of cellular aging and disease risk, but the genetics of telomere length control in humans is unclear. A genome-wide CRISPR screen has now identified a central role for thymidine nucleotide metabolism in the regulation of telomere length, which has implications for the diagnosis and treatment of disease.

A new study deciphers the origin and evolution of childhood neuroblastoma using genome sequencing data, mathematical models and statistical inference, showing how neuroblastoma evolution is an accurate predictor of outcome.

How histone modifications are reprogrammed through germline development in plants is poorly understood. We found that H3K27me3 and H3K4me3 are extensively reprogrammed throughout the development of Arabidopsis male gametophyte. This reprogramming leads to widespread chromatin bivalency and selective removal of H3K27me3 marks from key developmental regulators in sperm.

Quantifying whether different populations share similar effect sizes of common causal variants is vital to understand the genetic basis of disease and build better prediction models. A new study proposes a method leveraging admixture to estimate the correlation of causal genetic variants and finds they are largely similar across ancestry backgrounds.

We introduce molecular and cellular criteria — based on morphology, ploidy, CpG island methylation and immune infiltration — that improve the characterization of malignant pleural mesothelioma. These criteria reveal adaptation strategies that are adopted by tumor cells and offer new possibilities for classification and clinical management.

Current risk assessment and treatment strategies for venous thromboembolism (VTE) consider genetic factors only in a limited way. New work shows a more pervasive role of common variants in VTE risk, inspiring genetic predictors that surpass and complement individual clinical risk factors and monogenic thrombophilia testing.

The expression of murine endogenous retrovirus-L (MERVL) is transiently upregulated at the two-cell stage in mouse embryos, coinciding with zygotic genome activation and the acquisition of totipotency; however, its role in embryogenesis remains elusive. We show that nuclear expression of MERVL is required for accurate regulation of the host transcriptome and chromatin state during preimplantation development.

Interacting proteins commonly perform similar functions. A protein interaction network analysis of genes linked to 1,002 human traits identifies molecular mechanisms that are shared across traits and so-called pleiotropic biological processes, whose disruption might affect many aspects of human biology.

Multi-omic profiling of lesions at autopsy reveals a plethora of resistance mechanisms present within individual patients with ovarian cancer. This highlights the extreme challenge faced in treating end-stage disease and underscores the need for new methods of early detection and intervention.

A meta-analysis of harmonized human brain RNA-seq datasets creates expression quantitative trait locus (eQTL) maps for multiple ancestries and brain regions, predicts cell-type-dependent eQTLs and produces gene networks. This prioritizes genes for multiple brain-related diseases, serving as a promising step toward the identification of central nervous system (CNS) drug targets.

Using laser-capture microdissection and whole-genome sequencing of individual crypts, we characterized the landscape of somatic mutations in human small intestinal epithelium. Mutational signatures of APOBEC mutagenesis were found frequently and are probably due to the activity of APOBEC1, which is expressed at high levels in the small intestine.

Using a series of mouse mutants, we found that the Sox2 promoter does not require CTCF–cohesin loops to interact with distal enhancers. Surprisingly, mice with varying numbers of CTCF motifs in different positions showed that some distal enhancers can bypass boundaries that are created by CTCF–cohesin loops to ensure robust Sox2 expression.