Integration of variant annotations using deep set networks boosts rare variant association testing.

Rare genetic variants can have strong effects on phenotypes, yet accounting for rare variants in genetic analyses is statistically challenging due to the limited number of allele carriers and the burden of multiple testing. While rich variant annotations promise to enable well-powered rare variant association tests, methods integrating variant annotations in a data-driven manner are lacking. Here we propose deep rare variant association testing (DeepRVAT), a model based on set neural networks that learns a trait-agnostic gene impairment score from rare variant annotations and phenotypes, enabling both gene discovery and trait prediction. On 34 quantitative and 63 binary traits, using whole-exome-sequencing data from UK Biobank, we find that DeepRVAT yields substantial gains in gene discoveries and improved detection of individuals at high genetic risk. Finally, we demonstrate how DeepRVAT enables calibrated and computationally efficient rare variant tests at biobank scale, aiding the discovery of genetic risk factors for human disease traits.

scEpiAge: an age predictor highlighting single-cell ageing heterogeneity in mouse blood.

Ageing is the accumulation of changes and decline of function of organisms over time. The concept and biomarkers of biological age have been established, notably DNA methylation-based clocks. The emergence of single-cell DNA methylation profiling methods opens the possibility of studying the biological age of individual cells. Here, we generate a large single-cell DNA methylation and transcriptome dataset from mouse peripheral blood samples, spanning a broad range of ages. The number of genes expressed increases with age, but gene-specific changes are small. We next develop scEpiAge, a single-cell DNA methylation age predictor, which can accurately predict age in (very sparse) publicly available datasets, and also in single cells. DNA methylation age distribution is wider than technically expected, indicating epigenetic age heterogeneity and functional differences. Our work provides a foundation for single-cell and sparse data epigenetic age predictors, validates their functionality and highlights epigenetic heterogeneity during ageing.

Robust analysis of allele-specific copy number alterations from scRNA-seq data with XClone.

Somatic copy number alterations (CNAs) are major mutations that contribute to the development and progression of various cancers. Despite a few computational methods proposed to detect CNAs from single-cell transcriptomic data, the technical sparsity of such data makes it challenging to identify allele-specific CNAs, particularly in complex clonal structures. In this study, we present a statistical method, XClone, that strengthens the signals of read depth and allelic imbalance by effective smoothing on cell neighborhood and gene coordinate graphs to detect haplotype-aware CNAs from scRNA-seq data. By applying XClone to multiple datasets with challenging compositions, we demonstrated its ability to robustly detect different types of allele-specific CNAs and potentially indicate whole genome duplication, therefore enabling the discovery of corresponding subclones and the dissection of their phenotypic impacts.

Escape from X inactivation is directly modulated by levels of Xist non-coding RNA.

In placental females, one copy of the two X chromosomes is largely silenced during a narrow developmental time window, in a process mediated by the non-coding RNA Xist1. Here, we demonstrate that Xist can initiate X-chromosome inactivation (XCI) well beyond early embryogenesis. By modifying its endogenous level, we show that Xist has the capacity to actively silence genes that escape XCI both in neuronal progenitor cells (NPCs) and in vivo, in mouse embryos. We also show that Xist plays a direct role in eliminating TAD-like structures associated with clusters of escapee genes on the inactive X chromosome, and that this is dependent on Xist's XCI initiation partner, SPEN2. We further demonstrate that Xist's function in suppressing gene expression of escapees and topological domain formation is reversible for up to seven days post-induction, but that sustained Xist up-regulation leads to progressively irreversible silencing and CpG island DNA methylation of facultative escapees. Thus, the distinctive transcriptional and regulatory topologies of the silenced X chromosome is actively, directly - and reversibly - controlled by Xist RNA throughout life.

SpatialData: an open and universal data framework for spatial omics.

Spatially resolved omics technologies are transforming our understanding of biological tissues. However, the handling of uni- and multimodal spatial omics datasets remains a challenge owing to large data volumes, heterogeneity of data types and the lack of flexible, spatially aware data structures. Here we introduce SpatialData, a framework that establishes a unified and extensible multiplatform file-format, lazy representation of larger-than-memory data, transformations and alignment to common coordinate systems. SpatialData facilitates spatial annotations and cross-modal aggregation and analysis, the utility of which is illustrated in the context of multiple vignettes, including integrative analysis on a multimodal Xenium and Visium breast cancer study.

The dynamic genetic determinants of increased transcriptional divergence in spermatids.

Cis-genetic effects are key determinants of transcriptional divergence in discrete tissues and cell types. However, how cis- and trans-effects act across continuous trajectories of cellular differentiation in vivo is poorly understood. Here, we quantify allele-specific expression during spermatogenic differentiation at single-cell resolution in an F1 hybrid mouse system, allowing for the comprehensive characterisation of cis- and trans-genetic effects, including their dynamics across cellular differentiation. Collectively, almost half of the genes subject to genetic regulation show evidence for dynamic cis-effects that vary during differentiation. Our system also allows us to robustly identify dynamic trans-effects, which are less pervasive than cis-effects. In aggregate, genetic effects were strongest in round spermatids, which parallels their increased transcriptional divergence we identified between species. Our approach provides a comprehensive quantification of the variability of genetic effects in vivo, and demonstrates a widely applicable strategy to dissect the impact of regulatory variants on gene regulation in dynamic systems.

UNCAN.eu: Toward a European Federated Cancer Research Data Hub.

To enable a collective effort that generates a new level of UNderstanding CANcer (UNCAN.eu) [Cancer Discov (2022) 12 (11): OF1], the European Union supports the creation of a sustainable platform that connects cancer research across Member States. A workshop hosted in Heidelberg gathered European cancer experts to identify ongoing initiatives that may contribute to building this platform and discuss the governance and long-term evolution of a European Federated Cancer Data Hub.

Expression profiling of cerebrospinal fluid identifies dysregulated antiviral mechanisms in multiple sclerosis.

Despite the overwhelming evidence that multiple sclerosis is an autoimmune disease, relatively little is known about the precise nature of the immune dysregulation underlying the development of the disease. Reasoning that the CSF from patients might be enriched for cells relevant in pathogenesis, we have completed a high-resolution single-cell analysis of 96 732 CSF cells collected from 33 patients with multiple sclerosis (n = 48 675) and 48 patients with other neurological diseases (n = 48 057). Completing comprehensive cell type annotation, we identified a rare population of CD8+ T cells, characterized by the upregulation of inhibitory receptors, increased in patients with multiple sclerosis. Applying a Multi-Omics Factor Analysis to these single-cell data further revealed that activity in pathways responsible for controlling inflammatory and type 1 interferon responses are altered in multiple sclerosis in both T cells and myeloid cells. We also undertook a systematic search for expression quantitative trait loci in the CSF cells. Of particular interest were two expression quantitative trait loci in CD8+ T cells that were fine mapped to multiple sclerosis susceptibility variants in the viral control genes ZC3HAV1 (rs10271373) and IFITM2 (rs1059091). Further analysis suggests that these associations likely reflect genetic effects on RNA splicing and cell-type specific gene expression respectively. Collectively, our study suggests that alterations in viral control mechanisms might be important in the development of multiple sclerosis.

Aged intestinal stem cells propagate cell-intrinsic sources of inflammaging in mice.

Low-grade chronic inflammation is a hallmark of ageing, associated with impaired tissue function and disease development. However, how cell-intrinsic and -extrinsic factors collectively establish this phenotype, termed inflammaging, remains poorly understood. We addressed this question in the mouse intestinal epithelium, using mouse organoid cultures to dissect stem cell-intrinsic and -extrinsic sources of inflammaging. At the single-cell level, we found that inflammaging is established differently along the crypt-villus axis, with aged intestinal stem cells (ISCs) strongly upregulating major histocompatibility complex class II (MHC-II) genes. Importantly, the inflammaging phenotype was stably propagated by aged ISCs in organoid cultures and associated with increased chromatin accessibility at inflammation-associated loci in vivo and ex vivo, indicating cell-intrinsic inflammatory memory. Mechanistically, we show that the expression of inflammatory genes is dependent on STAT1 signaling. Together, our data identify that intestinal inflammaging in mice is promoted by a cell-intrinsic mechanism, stably propagated by ISCs, and associated with a disbalance in immune homeostasis.

Principles and challenges of modeling temporal and spatial omics data.

Studies with temporal or spatial resolution are crucial to understand the molecular dynamics and spatial dependencies underlying a biological process or system. With advances in high-throughput omic technologies, time- and space-resolved molecular measurements at scale are increasingly accessible, providing new opportunities to study the role of timing or structure in a wide range of biological questions. At the same time, analyses of the data being generated in the context of spatiotemporal studies entail new challenges that need to be considered, including the need to account for temporal and spatial dependencies and compare them across different scales, biological samples or conditions. In this Review, we provide an overview of common principles and challenges in the analysis of temporal and spatial omics data. We discuss statistical concepts to model temporal and spatial dependencies and highlight opportunities for adapting existing analysis methods to data with temporal and spatial dimensions.

A genome-wide association study of blood cell morphology identifies cellular proteins implicated in disease aetiology.

Blood cells contain functionally important intracellular structures, such as granules, critical to immunity and thrombosis. Quantitative variation in these structures has not been subjected previously to large-scale genetic analysis. We perform genome-wide association studies of 63 flow-cytometry derived cellular phenotypes-including cell-type specific measures of granularity, nucleic acid content and reactivity-in 41,515 participants in the INTERVAL study. We identify 2172 distinct variant-trait associations, including associations near genes coding for proteins in organelles implicated in inflammatory and thrombotic diseases. By integrating with epigenetic data we show that many intracellular structures are likely to be determined in immature precursor cells. By integrating with proteomic data we identify the transcription factor FOG2 as an early regulator of platelet formation and α-granularity. Finally, we show that colocalisation of our associations with disease risk signals can suggest aetiological cell-types-variants in IL2RA and ITGA4 respectively mirror the known effects of daclizumab in multiple sclerosis and vedolizumab in inflammatory bowel disease.

Mutation-Attention (MuAt): deep representation learning of somatic mutations for tumour typing and subtyping.

BACKGROUND: Cancer genome sequencing enables accurate classification of tumours and tumour subtypes. However, prediction performance is still limited using exome-only sequencing and for tumour types with low somatic mutation burden such as many paediatric tumours. Moreover, the ability to leverage deep representation learning in discovery of tumour entities remains unknown. METHODS: We introduce here Mutation-Attention (MuAt), a deep neural network to learn representations of simple and complex somatic alterations for prediction of tumour types and subtypes. In contrast to many previous methods, MuAt utilizes the attention mechanism on individual mutations instead of aggregated mutation counts. RESULTS: We trained MuAt models on 2587 whole cancer genomes (24 tumour types) from the Pan-Cancer Analysis of Whole Genomes (PCAWG) and 7352 cancer exomes (20 types) from the Cancer Genome Atlas (TCGA). MuAt achieved prediction accuracy of 89% for whole genomes and 64% for whole exomes, and a top-5 accuracy of 97% and 90%, respectively. MuAt models were found to be well-calibrated and perform well in three independent whole cancer genome cohorts with 10,361 tumours in total. We show MuAt to be able to learn clinically and biologically relevant tumour entities including acral melanoma, SHH-activated medulloblastoma, SPOP-associated prostate cancer, microsatellite instability, POLE proofreading deficiency, and MUTYH-associated pancreatic endocrine tumours without these tumour subtypes and subgroups being provided as training labels. Finally, scrunity of MuAt attention matrices revealed both ubiquitous and tumour-type specific patterns of simple and complex somatic mutations. CONCLUSIONS: Integrated representations of somatic alterations learnt by MuAt were able to accurately identify histological tumour types and identify tumour entities, with potential to impact precision cancer medicine.

Mapping interindividual dynamics of innate immune response at single-cell resolution.

Common genetic variants across individuals modulate the cellular response to pathogens and are implicated in diverse immune pathologies, yet how they dynamically alter the response upon infection is not well understood. Here, we triggered antiviral responses in human fibroblasts from 68 healthy donors, and profiled tens of thousands of cells using single-cell RNA-sequencing. We developed GASPACHO (GAuSsian Processes for Association mapping leveraging Cell HeterOgeneity), a statistical approach designed to identify nonlinear dynamic genetic effects across transcriptional trajectories of cells. This approach identified 1,275 expression quantitative trait loci (local false discovery rate 10%) that manifested during the responses, many of which were colocalized with susceptibility loci identified by genome-wide association studies of infectious and autoimmune diseases, including the OAS1 splicing quantitative trait locus in a COVID-19 susceptibility locus. In summary, our analytical approach provides a unique framework for delineation of the genetic variants that shape a wide spectrum of transcriptional responses at single-cell resolution.

Resolving therapy resistance mechanisms in multiple myeloma by multiomics subclone analysis.

Intratumor heterogeneity as a clinical challenge becomes most evident after several treatment lines, when multidrug-resistant subclones accumulate. To address this challenge, the characterization of resistance mechanisms at the subclonal level is key to identify common vulnerabilities. In this study, we integrate whole-genome sequencing, single-cell (sc) transcriptomics (scRNA sequencing), and chromatin accessibility (scATAC sequencing) together with mitochondrial DNA mutations to define subclonal architecture and evolution for longitudinal samples from 15 patients with relapsed or refractory multiple myeloma. We assess transcriptomic and epigenomic changes to resolve the multifactorial nature of therapy resistance and relate it to the parallel occurrence of different mechanisms: (1) preexisting epigenetic profiles of subclones associated with survival advantages, (2) converging phenotypic adaptation of genetically distinct subclones, and (3) subclone-specific interactions of myeloma and bone marrow microenvironment cells. Our study showcases how an integrative multiomics analysis can be applied to track and characterize distinct multidrug-resistant subclones over time for the identification of molecular targets against them.

pycoMeth: a toolbox for differential methylation testing from Nanopore methylation calls.

We present pycoMeth, a toolbox to store, manage and analyze DNA methylation calls from long-read sequencing data obtained using the Oxford Nanopore Technologies sequencing platform. Building on a novel, rapid-access, read-level and reference-anchored methylation storage format MetH5, we propose efficient algorithms for haplotype aware, multi-sample consensus segmentation and differential methylation testing. We show that MetH5 is more efficient than existing solutions for storing Oxford Nanopore Technologies methylation calls, and carry out benchmarking for pycoMeth segmentation and differential methylation testing, demonstrating increased performance and sensitivity compared to existing solutions designed for short-read methylation data.

Long-read sequencing of diagnosis and post-therapy medulloblastoma reveals complex rearrangement patterns and epigenetic signatures.

Cancer genomes harbor a broad spectrum of structural variants (SVs) driving tumorigenesis, a relevant subset of which escape discovery using short-read sequencing. We employed Oxford Nanopore Technologies (ONT) long-read sequencing in a paired diagnostic and post-therapy medulloblastoma to unravel the haplotype-resolved somatic genetic and epigenetic landscape. We assembled complex rearrangements, including a 1.55-Mbp chromothripsis event, and we uncover a complex SV pattern termed templated insertion (TI) thread, characterized by short (mostly <1 kb) insertions showing prevalent self-concatenation into highly amplified structures of up to 50 kbp in size. TI threads occur in 3% of cancers, with a prevalence up to 74% in liposarcoma, and frequent colocalization with chromothripsis. We also perform long-read-based methylome profiling and discover allele-specific methylation (ASM) effects, complex rearrangements exhibiting differential methylation, and differential promoter methylation in cancer-driver genes. Our study shows the advantage of long-read sequencing in the discovery and characterization of complex somatic rearrangements.

FISHFactor: a probabilistic factor model for spatial transcriptomics data with subcellular resolution.

MOTIVATION: Factor analysis is a widely used tool for unsupervised dimensionality reduction of high-throughput datasets in molecular biology, with recently proposed extensions designed specifically for spatial transcriptomics data. However, these methods expect (count) matrices as data input and are therefore not directly applicable to single molecule resolution data, which are in the form of coordinate lists annotated with genes and provide insight into subcellular spatial expression patterns. To address this, we here propose FISHFactor, a probabilistic factor model that combines the benefits of spatial, non-negative factor analysis with a Poisson point process likelihood to explicitly model and account for the nature of single molecule resolution data. In addition, FISHFactor shares information across a potentially large number of cells in a common weight matrix, allowing consistent interpretation of factors across cells and yielding improved latent variable estimates. RESULTS: We compare FISHFactor to existing methods that rely on aggregating information through spatial binning and cannot combine information from multiple cells and show that our method leads to more accurate results on simulated data. We show that our method is scalable and can be readily applied to large datasets. Finally, we demonstrate on a real dataset that FISHFactor is able to identify major subcellular expression patterns and spatial gene clusters in a data-driven manner. AVAILABILITY AND IMPLEMENTATION: The model implementation, data simulation and experiment scripts are available under https://www.github.com/bioFAM/FISHFactor.

Spatial multiomics map of trophoblast development in early pregnancy.

The relationship between the human placenta-the extraembryonic organ made by the fetus, and the decidua-the mucosal layer of the uterus, is essential to nurture and protect the fetus during pregnancy. Extravillous trophoblast cells (EVTs) derived from placental villi infiltrate the decidua, transforming the maternal arteries into high-conductance vessels1. Defects in trophoblast invasion and arterial transformation established during early pregnancy underlie common pregnancy disorders such as pre-eclampsia2. Here we have generated a spatially resolved multiomics single-cell atlas of the entire human maternal-fetal interface including the myometrium, which enables us to resolve the full trajectory of trophoblast differentiation. We have used this cellular map to infer the possible transcription factors mediating EVT invasion and show that they are preserved in in vitro models of EVT differentiation from primary trophoblast organoids3,4 and trophoblast stem cells5. We define the transcriptomes of the final cell states of trophoblast invasion: placental bed giant cells (fused multinucleated EVTs) and endovascular EVTs (which form plugs inside the maternal arteries). We predict the cell-cell communication events contributing to trophoblast invasion and placental bed giant cell formation, and model the dual role of interstitial EVTs and endovascular EVTs in mediating arterial transformation during early pregnancy. Together, our data provide a comprehensive analysis of postimplantation trophoblast differentiation that can be used to inform the design of experimental models of the human placenta in early pregnancy.