Zero-preserving imputation of single-cell RNA-seq data

scIALM: A method for sparse scRNA-seq expression matrix imputation using the Inexact Augmented Lagrange Multiplier with low error

Single-cell RNA sequencing (scRNA-seq) is a high-throughput sequencing technology that quantifies gene expression profiles of specific cell populations at the single-cell level, providing a foundation for studying cellular heterogeneity and patient pathological characteristics. It is effective for developmental, fertility, and disease studies. However, the cell-gene expression matrix of single-cell sequencing data is often sparse and contains numerous zero values. Some of the zero values derive from noise, where dropout noise has a large impact on downstream analysis. In this paper, we propose a method named scIALM for imputation recovery of sparse single-cell RNA data expression matrices, which employs the Inexact Augmented Lagrange Multiplier method to use sparse but clean (accurate) data to recover unknown entries in the matrix. We perform experimental analysis on four datasets, calling the expression matrix after Quality Control (QC) as the original matrix, and comparing the performance of scIALM with six other methods using mean squared error (MSE), mean absolute error (MAE), Pearson correlation coefficient (PCC), and cosine similarity (CS). Our results demonstrate that scIALM accurately recovers the original data of the matrix with an error of 10e-4, and the mean value of the four metrics reaches 4.5072 (MSE), 0.765 (MAE), 0.8701 (PCC), 0.8896 (CS). In addition, at 10%-50% random masking noise, scIALM is the least sensitive to the masking ratio. For downstream analysis, this study uses adjusted rand index (ARI) and normalized mutual information (NMI) to evaluate the clustering effect, and the results are improved on three datasets containing real cluster labels.

Lineage specification in glioblastoma is regulated by METTL7B

Glioblastomas are the most common malignant brain tumors in adults; they are highly aggressive and heterogeneous and show a high degree of plasticity. Here, we show that methyltransferase-like 7B (METTL7B) is an essential regulator of lineage specification in glioblastoma, with an impact on both tumor size and invasiveness. Single-cell transcriptomic analysis of these tumors and of cerebral organoids derived from expanded potential stem cells overexpressing METTL7B reveal a regulatory role for the gene in the neural stem cell-to-astrocyte differentiation trajectory. Mechanistically, METTL7B downregulates the expression of key neuronal differentiation players, including SALL2, via post-translational modifications of histone marks.

Impeller: a path-based heterogeneous graph learning method for spatial transcriptomic data imputation

MOTIVATION

Recent advances in spatial transcriptomics allow spatially resolved gene expression measurements with cellular or even sub-cellular resolution, directly characterizing the complex spatiotemporal gene expression landscape and cell-to-cell interactions in their native microenvironments. Due to technology limitations, most spatial transcriptomic technologies still yield incomplete expression measurements with excessive missing values. Therefore, gene imputation is critical to filling in missing data, enhancing resolution, and improving overall interpretability. However, existing methods either require additional matched single-cell RNA-seq data, which is rarely available, or ignore spatial proximity or expression similarity information.

RESULTS

To address these issues, we introduce Impeller, a path-based heterogeneous graph learning method for spatial transcriptomic data imputation. Impeller has two unique characteristics distinct from existing approaches. First, it builds a heterogeneous graph with two types of edges representing spatial proximity and expression similarity. Therefore, Impeller can simultaneously model smooth gene expression changes across spatial dimensions and capture similar gene expression signatures of faraway cells from the same type. Moreover, Impeller incorporates both short- and long-range cell-to-cell interactions (e.g. via paracrine and endocrine) by stacking multiple GNN layers. We use a learnable path operator in Impeller to avoid the over-smoothing issue of the traditional Laplacian matrices. Extensive experiments on diverse datasets from three popular platforms and two species demonstrate the superiority of Impeller over various state-of-the-art imputation methods.

AVAILABILITY AND IMPLEMENTATION

The code and preprocessed data used in this study are available at https://github.com/aicb-ZhangLabs/Impeller and https://zenodo.org/records/11212604.

NK Cell-Monocyte Cross-talk Underlies NK Cell Activation in Severe COVID-19

NK cells in the peripheral blood of severe COVID-19 patients exhibit a unique profile characterized by activation and dysfunction. Previous studies have identified soluble factors, including type I IFN and TGF-β, that underlie this dysregulation. However, the role of cell-cell interactions in modulating NK cell function during COVID-19 remains unclear. To address this question, we combined cell-cell communication analysis on existing single-cell RNA sequencing data with in vitro primary cell coculture experiments to dissect the mechanisms underlying NK cell dysfunction in COVID-19. We found that NK cells are predicted to interact most strongly with monocytes and that this occurs via both soluble factors and direct interactions. To validate these findings, we performed in vitro cocultures in which NK cells from healthy human donors were incubated with monocytes from COVID-19+ or healthy donors. Coculture of healthy NK cells with monocytes from COVID-19 patients recapitulated aspects of the NK cell phenotype observed in severe COVID-19, including decreased expression of NKG2D, increased expression of activation markers, and increased proliferation. When these experiments were performed in a Transwell setting, we found that only CD56bright CD16- NK cells were activated in the presence of severe COVID-19 patient monocytes. O-link analysis of supernatants from Transwell cocultures revealed that cultures containing severe COVID-19 patient monocytes had significantly elevated levels of proinflammatory cytokines and chemokines, as well as TGF-β. Collectively, these results demonstrate that interactions between NK cells and monocytes in the peripheral blood of COVID-19 patients contribute to NK cell activation and dysfunction in severe COVID-19.

IL-4 Licenses B Cell Activation Through Cholesterol Synthesis

Lymphocyte activation involves a transition from quiescence and associated catabolic metabolism to a metabolic state with noted similarities to cancer cells such as heavy reliance on aerobic glycolysis for energy demands and increased nutrient requirements for biomass accumulation and cell division 1-3 . Following antigen receptor ligation, lymphocytes require spatiotemporally distinct "second signals". These include costimulatory receptor or cytokine signaling, which engage discrete programs that often involve remodeling of organelles and increased nutrient uptake or synthesis to meet changing biochemical demands 4-6 . One such signaling molecule, IL-4, is a highly pleiotropic cytokine that was first identified as a B cell co-mitogen over 30 years ago 7 . However, how IL-4 signaling mechanistically supports B cell proliferation is incompletely understood. Here, using single cell RNA sequencing we find that the cholesterol biosynthetic program is transcriptionally upregulated following IL-4 signaling during the early B cell response to influenza virus infection, and is required for B cell activation in vivo . By limiting lipid availability in vitro , we determine cholesterol to be essential for B cells to expand their endoplasmic reticulum, progress through cell cycle, and proliferate. In sum, we demonstrate that the well-known ability of IL-4 to act as a B cell growth factor is through a previously unknown rewiring of specific lipid anabolic programs, relieving sensitivity of cells to environmental nutrient availability.

Data normalization for addressing the challenges in the analysis of single-cell transcriptomic datasets

BACKGROUND

Normalization is a critical step in the analysis of single-cell RNA-sequencing (scRNA-seq) datasets. Its main goal is to make gene counts comparable within and between cells. To do so, normalization methods must account for technical and biological variability. Numerous normalization methods have been developed addressing different sources of dispersion and making specific assumptions about the count data.

MAIN BODY

The selection of a normalization method has a direct impact on downstream analysis, for example differential gene expression and cluster identification. Thus, the objective of this review is to guide the reader in making an informed decision on the most appropriate normalization method to use. To this aim, we first give an overview of the different single cell sequencing platforms and methods commonly used including isolation and library preparation protocols. Next, we discuss the inherent sources of variability of scRNA-seq datasets. We describe the categories of normalization methods and include examples of each. We also delineate imputation and batch-effect correction methods. Furthermore, we describe data-driven metrics commonly used to evaluate the performance of normalization methods. We also discuss common scRNA-seq methods and toolkits used for integrated data analysis.

CONCLUSIONS

According to the correction performed, normalization methods can be broadly classified as within and between-sample algorithms. Moreover, with respect to the mathematical model used, normalization methods can further be classified into: global scaling methods, generalized linear models, mixed methods, and machine learning-based methods. Each of these methods depict pros and cons and make different statistical assumptions. However, there is no better performing normalization method. Instead, metrics such as silhouette width, K-nearest neighbor batch-effect test, or Highly Variable Genes are recommended to assess the performance of normalization methods.

scLENS: data-driven signal detection for unbiased scRNA-seq data analysis

High dimensionality and noise have limited the new biological insights that can be discovered in scRNA-seq data. While dimensionality reduction tools have been developed to extract biological signals from the data, they often require manual determination of signal dimension, introducing user bias. Furthermore, a common data preprocessing method, log normalization, can unintentionally distort signals in the data. Here, we develop scLENS, a dimensionality reduction tool that circumvents the long-standing issues of signal distortion and manual input. Specifically, we identify the primary cause of signal distortion during log normalization and effectively address it by uniformizing cell vector lengths with L2 normalization. Furthermore, we utilize random matrix theory-based noise filtering and a signal robustness test to enable data-driven determination of the threshold for signal dimensions. Our method outperforms 11 widely used dimensionality reduction tools and performs particularly well for challenging scRNA-seq datasets with high sparsity and variability. To facilitate the use of scLENS, we provide a user-friendly package that automates accurate signal detection of scRNA-seq data without manual time-consuming tuning.

Cancer-associated fibroblast-derived Dickkopf-1 suppresses NK cell cytotoxicity in breast cancer

Breast cancer is poorly immunogenic, hence able to evade T cell recognition and respond poorly to immune checkpoint blockade. Breast cancer cells can also evade NK cell-mediated immune surveillance, but the mechanism remains enigmatic. Dickkopf-1 (DKK1) is a Wnt/b-catenin inhibitor, whose levels are increased in breast cancer patients and correlate with reduced overall survival. DKK1 is expressed by cancer-associated fibroblasts (CAFs) in orthotopic breast tumors and patient samples, and at higher levels by bone cells. While bone-derived DKK1 contributes to the systemic elevation of DKK1 in tumor-bearing mice, CAFs represent the primary source of DKK1 at the tumor site. Systemic or bone-specific DKK1 targeting reduces primary tumor growth. Intriguingly, specific deletion of CAF-derived DKK1 also limits breast cancer progression, regardless of its elevated levels in circulation and in the bone. DKK1 does not support tumor proliferation directly but rather suppresses the activation and tumoricidal activity of NK cells. Importantly, increased DKK1 levels and reduced number of cytotoxic NK cells are detected in breast cancer patients with progressive bone metastases compared to those with stable disease. Our findings indicate that DKK1 creates a tumor-supporting environment through the suppression of NK cells in breast cancer.

scButterfly: a versatile single-cell cross-modality translation method via dual-aligned variational autoencoders

Recent advancements for simultaneously profiling multi-omics modalities within individual cells have enabled the interrogation of cellular heterogeneity and molecular hierarchy. However, technical limitations lead to highly noisy multi-modal data and substantial costs. Although computational methods have been proposed to translate single-cell data across modalities, broad applications of the methods still remain impeded by formidable challenges. Here, we propose scButterfly, a versatile single-cell cross-modality translation method based on dual-aligned variational autoencoders and data augmentation schemes. With comprehensive experiments on multiple datasets, we provide compelling evidence of scButterfly's superiority over baseline methods in preserving cellular heterogeneity while translating datasets of various contexts and in revealing cell type-specific biological insights. Besides, we demonstrate the extensive applications of scButterfly for integrative multi-omics analysis of single-modality data, data enhancement of poor-quality single-cell multi-omics, and automatic cell type annotation of scATAC-seq data. Moreover, scButterfly can be generalized to unpaired data training, perturbation-response analysis, and consecutive translation.

Distribution-Agnostic Deep Learning Enables Accurate Single-Cell Data Recovery and Transcriptional Regulation Interpretation

Single-cell RNA sequencing (scRNA-seq) is a robust method for studying gene expression at the single-cell level, but accurately quantifying genetic material is often hindered by limited mRNA capture, resulting in many missing expression values. Existing imputation methods rely on strict data assumptions, limiting their broader application, and lack reliable supervision, leading to biased signal recovery. To address these challenges, authors developed Bis, a distribution-agnostic deep learning model for accurately recovering missing sing-cell gene expression from multiple platforms. Bis is an optimal transport-based autoencoder model that can capture the intricate distribution of scRNA-seq data while addressing the characteristic sparsity by regularizing the cellular embedding space. Additionally, they propose a module using bulk RNA-seq data to guide reconstruction and ensure expression consistency. Experimental results show Bis outperforms other models across simulated and real datasets, showcasing superiority in various downstream analyses including batch effect removal, clustering, differential expression analysis, and trajectory inference. Moreover, Bis successfully restores gene expression levels in rare cell subsets in a tumor-matched peripheral blood dataset, revealing developmental characteristics of cytokine-induced natural killer cells within a head and neck squamous cell carcinoma microenvironment.

Reconstruction Set Test (RESET): A computationally efficient method for single sample gene set testing based on randomized reduced rank reconstruction error

We have developed a new, and analytically novel, single sample gene set testing method called Reconstruction Set Test (RESET). RESET quantifies gene set importance based on the ability of set genes to reconstruct values for all measured genes. RESET is realized using a computationally efficient randomized reduced rank reconstruction algorithm (available via the RESET R package on CRAN) that can effectively detect patterns of differential abundance and differential correlation for self-contained and competitive scenarios. As demonstrated using real and simulated scRNA-seq data, RESET provides superior performance at a lower computational cost relative to other single sample approaches.

Single-cell RNA sequencing data imputation using bi-level feature propagation

Single-cell RNA sequencing (scRNA-seq) enables the exploration of cellular heterogeneity by analyzing gene expression profiles in complex tissues. However, scRNA-seq data often suffer from technical noise, dropout events and sparsity, hindering downstream analyses. Although existing works attempt to mitigate these issues by utilizing graph structures for data denoising, they involve the risk of propagating noise and fall short of fully leveraging the inherent data relationships, relying mainly on one of cell-cell or gene-gene associations and graphs constructed by initial noisy data. To this end, this study presents single-cell bilevel feature propagation (scBFP), two-step graph-based feature propagation method. It initially imputes zero values using non-zero values, ensuring that the imputation process does not affect the non-zero values due to dropout. Subsequently, it denoises the entire dataset by leveraging gene-gene and cell-cell relationships in the respective steps. Extensive experimental results on scRNA-seq data demonstrate the effectiveness of scBFP in various downstream tasks, uncovering valuable biological insights.

scINRB: single-cell gene expression imputation with network regularization and bulk RNA-seq data

Single-cell RNA sequencing (scRNA-seq) facilitates the study of cell type heterogeneity and the construction of cell atlas. However, due to its limitations, many genes may be detected to have zero expressions, i.e. dropout events, leading to bias in downstream analyses and hindering the identification and characterization of cell types and cell functions. Although many imputation methods have been developed, their performances are generally lower than expected across different kinds and dimensions of data and application scenarios. Therefore, developing an accurate and robust single-cell gene expression data imputation method is still essential. Considering to maintain the original cell-cell and gene-gene correlations and leverage bulk RNA sequencing (bulk RNA-seq) data information, we propose scINRB, a single-cell gene expression imputation method with network regularization and bulk RNA-seq data. scINRB adopts network-regularized non-negative matrix factorization to ensure that the imputed data maintains the cell-cell and gene-gene similarities and also approaches the gene average expression calculated from bulk RNA-seq data. To evaluate the performance, we test scINRB on simulated and experimental datasets and compare it with other commonly used imputation methods. The results show that scINRB recovers gene expression accurately even in the case of high dropout rates and dimensions, preserves cell-cell and gene-gene similarities and improves various downstream analyses including visualization, clustering and trajectory inference.

Recent advances in differential expression analysis for single-cell RNA-seq and spatially resolved transcriptomic studies

Differential expression (DE) analysis is a necessary step in the analysis of single-cell RNA sequencing (scRNA-seq) and spatially resolved transcriptomics (SRT) data. Unlike traditional bulk RNA-seq, DE analysis for scRNA-seq or SRT data has unique characteristics that may contribute to the difficulty of detecting DE genes. However, the plethora of DE tools that work with various assumptions makes it difficult to choose an appropriate one. Furthermore, a comprehensive review on detecting DE genes for scRNA-seq data or SRT data from multi-condition, multi-sample experimental designs is lacking. To bridge such a gap, here, we first focus on the challenges of DE detection, then highlight potential opportunities that facilitate further progress in scRNA-seq or SRT analysis, and finally provide insights and guidance in selecting appropriate DE tools or developing new computational DE methods.

Primary nasal viral infection rewires the tissue-scale memory response

The nasal mucosa is frequently the initial site of respiratory viral infection, replication, and transmission. Recent work has started to clarify the independent responses of epithelial, myeloid, and lymphoid cells to viral infection in the nasal mucosa, but their spatiotemporal coordination and relative contributions remain unclear. Furthermore, understanding whether and how primary infection shapes tissue-scale memory responses to secondary challenge is critical for the rational design of nasal-targeting therapeutics and vaccines. Here, we generated a single-cell RNA-sequencing (scRNA-seq) atlas of the murine nasal mucosa sampling three distinct regions before and during primary and secondary influenza infection. Primary infection was largely restricted to respiratory mucosa and induced stepwise changes in cell type, subset, and state composition over time. Type I Interferon (IFN)-responsive neutrophils appeared 2 days post infection (dpi) and preceded transient IFN-responsive/cycling epithelial cell responses 5 dpi, which coincided with broader antiviral monocyte and NK cell accumulation. By 8 dpi, monocyte-derived macrophages (MDMs) expressing Cxcl9 and Cxcl16 arose alongside effector cytotoxic CD8 and Ifng-expressing CD4 T cells. Following viral clearance (14 dpi), rare, previously undescribed Krt13+ nasal immune-interacting floor epithelial (KNIIFE) cells expressing multiple genes with immune communication potential increased concurrently with tissue-resident memory T (TRM)-like cells and early IgG+/IgA+ plasmablasts. Proportionality analysis coupled with cell-cell communication inference, alongside validation by in situ microscopy, underscored the CXCL16-CXCR6 signaling axis between MDMs and effector CD8 T cells 8dpi and KNIIFE cells and TRM cells 14 dpi. Secondary influenza challenge with a homologous or heterologous strain administered 60 dpi induced an accelerated and coordinated myeloid and lymphoid response without epithelial proliferation, illustrating how tissue-scale memory to natural infection engages both myeloid and lymphoid cells to reduce epithelial regenerative burden. Together, this atlas serves as a reference for viral infection in the upper respiratory tract and highlights the efficacy of local coordinated memory responses upon rechallenge.

Neutrophil profiling illuminates anti-tumor antigen-presenting potency

Neutrophils, the most abundant and efficient defenders against pathogens, exert opposing functions across cancer types. However, given their short half-life, it remains challenging to explore how neutrophils adopt specific fates in cancer. Here, we generated and integrated single-cell neutrophil transcriptomes from 17 cancer types (225 samples from 143 patients). Neutrophils exhibited extraordinary complexity, with 10 distinct states including inflammation, angiogenesis, and antigen presentation. Notably, the antigen-presenting program was associated with favorable survival in most cancers and could be evoked by leucine metabolism and subsequent histone H3K27ac modification. These neutrophils could further invoke both (neo)antigen-specific and antigen-independent T cell responses. Neutrophil delivery or a leucine diet fine-tuned the immune balance to enhance anti-PD-1 therapy in various murine cancer models. In summary, these data not only indicate the neutrophil divergence across cancers but also suggest therapeutic opportunities such as antigen-presenting neutrophil delivery.

scBiG for representation learning of single-cell gene expression data based on bipartite graph embedding

Analyzing single-cell RNA sequencing (scRNA-seq) data remains a challenge due to its high dimensionality, sparsity and technical noise. Recognizing the benefits of dimensionality reduction in simplifying complexity and enhancing the signal-to-noise ratio, we introduce scBiG, a novel graph node embedding method designed for representation learning in scRNA-seq data. scBiG establishes a bipartite graph connecting cells and expressed genes, and then constructs a multilayer graph convolutional network to learn cell and gene embeddings. Through a series of extensive experiments, we demonstrate that scBiG surpasses commonly used dimensionality reduction techniques in various analytical tasks. Downstream tasks encompass unsupervised cell clustering, cell trajectory inference, gene expression reconstruction and gene co-expression analysis. Additionally, scBiG exhibits notable computational efficiency and scalability. In summary, scBiG offers a useful graph neural network framework for representation learning in scRNA-seq data, empowering a diverse array of downstream analyses.

IL-10 constrains sphingolipid metabolism to limit inflammation

Interleukin-10 (IL-10) is a key anti-inflammatory cytokine that can limit immune cell activation and cytokine production in innate immune cell types1. Loss of IL-10 signalling results in life-threatening inflammatory bowel disease in humans and mice-however, the exact mechanism by which IL-10 signalling subdues inflammation remains unclear2-5. Here we find that increased saturated very long chain (VLC) ceramides are critical for the heightened inflammatory gene expression that is a hallmark of IL-10 deficiency. Accordingly, genetic deletion of ceramide synthase 2 (encoded by Cers2), the enzyme responsible for VLC ceramide production, limited the exacerbated inflammatory gene expression programme associated with IL-10 deficiency both in vitro and in vivo. The accumulation of saturated VLC ceramides was regulated by a decrease in metabolic flux through the de novo mono-unsaturated fatty acid synthesis pathway. Restoring mono-unsaturated fatty acid availability to cells deficient in IL-10 signalling limited saturated VLC ceramide production and the associated inflammation. Mechanistically, we find that persistent inflammation mediated by VLC ceramides is largely dependent on sustained activity of REL, an immuno-modulatory transcription factor. Together, these data indicate that an IL-10-driven fatty acid desaturation programme rewires VLC ceramide accumulation and aberrant activation of REL. These studies support the idea that fatty acid homeostasis in innate immune cells serves as a key regulatory node to control pathologic inflammation and suggests that 'metabolic correction' of VLC homeostasis could be an important strategy to normalize dysregulated inflammation caused by the absence of IL-10.

Comparative analysis of cell-cell communication at single-cell resolution

Inference of cell-cell communication from single-cell RNA sequencing data is a powerful technique to uncover intercellular communication pathways, yet existing methods perform this analysis at the level of the cell type or cluster, discarding single-cell-level information. Here we present Scriabin, a flexible and scalable framework for comparative analysis of cell-cell communication at single-cell resolution that is performed without cell aggregation or downsampling. We use multiple published atlas-scale datasets, genetic perturbation screens and direct experimental validation to show that Scriabin accurately recovers expected cell-cell communication edges and identifies communication networks that can be obscured by agglomerative methods. Additionally, we use spatial transcriptomic data to show that Scriabin can uncover spatial features of interaction from dissociated data alone. Finally, we demonstrate applications to longitudinal datasets to follow communication pathways operating between timepoints. Our approach represents a broadly applicable strategy to reveal the full structure of niche-phenotype relationships in health and disease.

Drug target prediction through deep learning functional representation of gene signatures

Many machine learning applications in bioinformatics currently rely on matching gene identities when analyzing input gene signatures and fail to take advantage of preexisting knowledge about gene functions. To further enable comparative analysis of OMICS datasets, including target deconvolution and mechanism of action studies, we develop an approach that represents gene signatures projected onto their biological functions, instead of their identities, similar to how the word2vec technique works in natural language processing. We develop the Functional Representation of Gene Signatures (FRoGS) approach by training a deep learning model and demonstrate that its application to the Broad Institute's L1000 datasets results in more effective compound-target predictions than models based on gene identities alone. By integrating additional pharmacological activity data sources, FRoGS significantly increases the number of high-quality compound-target predictions relative to existing approaches, many of which are supported by in silico and/or experimental evidence. These results underscore the general utility of FRoGS in machine learning-based bioinformatics applications. Prediction networks pre-equipped with the knowledge of gene functions may help uncover new relationships among gene signatures acquired by large-scale OMICs studies on compounds, cell types, disease models, and patient cohorts.

Circulating Tumor Reactive KIR+CD8+ T cells Suppress Anti-Tumor Immunity in Patients with Melanoma

Effective anti-tumor immunity is largely driven by cytotoxic CD8+ T cells that can specifically recognize tumor antigens. However, the factors which ultimately dictate successful tumor rejection remain poorly understood. Here we identify a subpopulation of CD8+ T cells which are tumor antigen-specific in patients with melanoma but resemble KIR+CD8+ T cells with a regulatory function (Tregs). These tumor antigen-specific KIR+CD8+ T cells are detectable in both the tumor and the blood, and higher levels of this population are associated with worse overall survival. Our findings therefore suggest that KIR+CD8+ Tregs are tumor antigen-specific but uniquely suppress anti-tumor immunity in patients with melanoma.

cnnImpute: missing value recovery for single cell RNA sequencing data

The advent of single-cell RNA sequencing (scRNA-seq) technology has revolutionized our ability to explore cellular diversity and unravel the complexities of intricate diseases. However, due to the inherently low signal-to-noise ratio and the presence of an excessive number of missing values, scRNA-seq data analysis encounters unique challenges. Here, we present cnnImpute, a novel convolutional neural network (CNN) based method designed to address the issue of missing data in scRNA-seq. Our approach starts by estimating missing probabilities, followed by constructing a CNN-based model to recover expression values with a high likelihood of being missing. Through comprehensive evaluations, cnnImpute demonstrates its effectiveness in accurately imputing missing values while preserving the integrity of cell clusters in scRNA-seq data analysis. It achieved superior performance in various benchmarking experiments. cnnImpute offers an accurate and scalable method for recovering missing values, providing a useful resource for scRNA-seq data analysis.

Targeted checkpoint control of B cells undergoing positive selection in germinal centers by follicular regulatory T cells

Follicular regulatory T cells (Tfr) can play opposite roles in the regulation of germinal center (GC) responses. Depending on the studies, Tfr suppress or support GC and B cell affinity maturation. However, which factors determine positive vs. negative effects of Tfr on the GC B cell is unclear. In this study, we show that GC centrocytes that express MYC up-regulate expression of CCL3 chemokine that is needed for both the positive and negative regulation of GC B cells by Tfr. B cell-intrinsic expression of CCL3 contributes to Tfr-dependent positive selection of foreign Ag-specific GC B cells. At the same time, expression of CCL3 is critical for direct Tfr-mediated suppression of GC B cells that acquire cognate to Tfr nuclear proteins. Our study suggests that CCR5 and CCR1 receptors promote Tfr migration to CCL3 and highlights Ccr5 expression on the Tfr subset that expresses Il10. Based on our findings and previous studies, we suggest a model of chemotactically targeted checkpoint control of B cells undergoing positive selection in GCs by Tfr, where Tfr directly probe and license foreign antigen-specific B cells to complete their positive selection in GCs but, at the same time, suppress GC B cells that present self-antigens cognate to Tfr.

ADGRE5-centered Tsurv model in T cells recognizes responders to neoadjuvant cancer immunotherapy

Background

Neoadjuvant immunotherapy with anti-programmed death-1 (neo-antiPD1) has revolutionized perioperative methods for improvement of overall survival (OS), while approaches for major pathologic response patients' (MPR) recognition along with methods for overcoming non-MPR resistance are still in urgent need.

Methods

We utilized and integrated publicly-available immune checkpoint inhibitors regimens (ICIs) single-cell (sc) data as the discovery datasets, and innovatively developed a cell-communication analysis pipeline, along with a VIPER-based-SCENIC process, to thoroughly dissect MPR-responding subsets. Besides, we further employed our own non-small cell lung cancer (NSCLC) ICIs cohort's sc data for validation in-silico. Afterward, we resorted to ICIs-resistant murine models developed by us with multimodal investigation, including bulk-RNA-sequencing, Chip-sequencing and high-dimensional cytometry by time of flight (CYTOF) to consolidate our findings in-vivo. To comprehensively explore mechanisms, we adopted 3D ex-vivo hydrogel models for analysis. Furthermore, we constructed an ADGRE5-centered Tsurv model from our discovery dataset by machine learning (ML) algorithms for a wide range of tumor types (NSCLC, melanoma, urothelial cancer, etc.) and verified it in peripheral blood mononuclear cells (PBMCs) sc datasets.

Results

Through a meta-analysis of multimodal sequential sc sequencing data from pre-ICIs and post-ICIs, we identified an MPR-expanding T cells meta-cluster (MPR-E) in the tumor microenvironment (TME), characterized by a stem-like CD8+ T cluster (survT) with STAT5-ADGRE5 axis enhancement compared to non-MPR or pre-ICIs TME. Through multi-omics analysis of murine TME, we further confirmed the existence of survT with silenced function and immune checkpoints (ICs) in MPR-E. After verification of the STAT5-ADGRE5 axis of survT in independent ICIs cohorts, an ADGRE5-centered Tsurv model was then developed through ML for identification of MPR patients pre-ICIs and post-ICIs, both in TME and PBMCs, which was further verified in pan-cancer immunotherapy cohorts. Mechanistically, we unveiled ICIs stimulated ADGRE5 upregulation in a STAT5-IL32 dependent manner in a 3D ex-vivo system (3D-HYGTIC) developed by us previously, which marked Tsurv with better survival flexibility, enhanced stemness and potential cytotoxicity within TME.

Conclusion

Our research provides insights into mechanisms underlying MPR in neo-antiPD1 and a well-performed model for the identification of non-MPR.

mRNA-based precision targeting of neoantigens and tumor-associated antigens in malignant brain tumors

BACKGROUND

Despite advancements in the successful use of immunotherapy in treating a variety of solid tumors, applications in treating brain tumors have lagged considerably. This is due, at least in part, to the lack of well-characterized antigens expressed within brain tumors that can mediate tumor rejection; the low mutational burden of these tumors that limits the abundance of targetable neoantigens; and the immunologically "cold" tumor microenvironment that hampers the generation of sustained and productive immunologic responses. The field of mRNA-based therapeutics has experienced a boon following the universal approval of COVID-19 mRNA vaccines. mRNA-based immunotherapeutics have also garnered widespread interest for their potential to revolutionize cancer treatment. In this study, we developed a novel and scalable approach for the production of personalized mRNA-based therapeutics that target multiple tumor rejection antigens in a single therapy for the treatment of refractory brain tumors.

METHODS

Tumor-specific neoantigens and aberrantly overexpressed tumor-associated antigens were identified for glioblastoma and medulloblastoma tumors using our cancer immunogenomics pipeline called Open Reading Frame Antigen Network (O.R.A.N). Personalized tumor antigen-specific mRNA vaccine was developed for each individual tumor model using selective gene capture and enrichment strategy. The immunogenicity and efficacy of the personalized mRNA vaccines was evaluated in combination with anti-PD-1 immune checkpoint blockade therapy or adoptive cellular therapy with ex vivo expanded tumor antigen-specific lymphocytes in highly aggressive murine GBM models.

RESULTS

Our results demonstrate the effectiveness of the antigen-specific mRNA vaccines in eliciting robust anti-tumor immune responses in GBM hosts. Our findings substantiate an increase in tumor-infiltrating lymphocytes characterized by enhanced effector function, both intratumorally and systemically, after antigen-specific mRNA-directed immunotherapy, resulting in a favorable shift in the tumor microenvironment from immunologically cold to hot. Capacity to generate personalized mRNA vaccines targeting human GBM antigens was also demonstrated.

CONCLUSIONS

We have established a personalized and customizable mRNA-therapeutic approach that effectively targets a plurality of tumor antigens and demonstrated potent anti-tumor response in preclinical brain tumor models. This platform mRNA technology uniquely addresses the challenge of tumor heterogeneity and low antigen burden, two key deficiencies in targeting the classically immunotherapy-resistant CNS malignancies, and possibly other cold tumor types.

Organ Boundary Circuits Regulate Sox9+ Alveolar Tuft Cells During Post-Pneumonectomy Lung Regeneration

Tissue homeostasis is controlled by cellular circuits governing cell growth, organization, and differentation. In this study we identify previously undescribed cell-to-cell communication that mediates information flow from mechanosensitive pleural mesothelial cells to alveolar-resident stem-like tuft cells in the lung. We find mesothelial cells to express a combination of mechanotransduction genes and lineage-restricted ligands which makes them uniquely capable of responding to tissue tension and producing paracrine cues acting on parenchymal populations. In parallel, we describe a large population of stem-like alveolar tuft cells that express the endodermal stem cell markers Sox9 and Lgr5 and a receptor profile making them uniquely sensitive to cues produced by pleural Mesothelium. We hypothesized that crosstalk from mesothelial cells to alveolar tuft cells might be central to the regulation of post-penumonectomy lung regeneration. Following pneumonectomy, we find that mesothelial cells display radically altered phenotype and ligand expression, in a pattern that closely tracks with parenchymal epithelial proliferation and alveolar tissue growth. During an initial pro-inflammatory stage of tissue regeneration, Mesothelium promotes epithelial proliferation via WNT ligand secretion, orchestrates an increase in microvascular permeability, and encourages immune extravasation via chemokine secretion. This stage is followed first by a tissue remodeling period, characterized by angiogenesis and BMP pathway sensitization, and then a stable return to homeostasis. Coupled with key changes in parenchymal structure and matrix production, the cumulative effect is a now larger organ including newly-grown, fully-functional tissue parenchyma. This study paints Mesothelial cells as a key orchestrating cell type that defines the boundary of the lung and exerts critical influence over the tissue-level signaling state regulating resident stem cell populations. The cellular circuits unearthed here suggest that human lung regeneration might be inducible through well-engineered approaches targeting the induction of tissue regeneration and safe return to homeostasis.

scCAN: Clustering With Adaptive Neighbor-Based Imputation Method for Single-Cell RNA-Seq Data

Single-cell RNA sequencing (scRNA-seq) is widely used to study cellular heterogeneity in different samples. However, due to technical deficiencies, dropout events often result in zero gene expression values in the gene expression matrix. In this paper, we propose a new imputation method called scCAN, based on adaptive neighborhood clustering, to estimate the zero value of dropouts. Our method continuously updates cell-cell similarity information by simultaneously learning similarity relationships, clustering structures, and imposing new rank constraints on the Laplacian matrix of the similarity matrix, improving the imputation of dropout zero values. To evaluate the performance of this method, we used four simulated and eight real scRNA-seq data for downstream analyses, including cell clustering, recovered gene expression, and reconstructed cell trajectories. Our method improves the performance of the downstream analysis and is better than other imputation methods.

From Cell States to Cell Fates: Control of Cell State Transitions

We examine the coordinated behavior of thousands of genes in cell fate transitions through genome expression as an integrated dynamical system using the concepts of self-organized criticality and coherent stochastic behavior. To quantify the effects of the collective behavior of genes, we adopted the flux balance approach and developed it in a new tool termed expression flux analysis (EFA). Here we describe this tool and demonstrate how its application to specific experimental genome-wide expression data provides new insights into the dynamics of the cell-fate transitions. Particularly, we show that in cell fate change, specific stochastic perturbations can spread over the entire system to guide distinct cell fate transitions through switching cyclic flux flow in the genome engine. Utilization of EFA enables us to elucidate a unified genomic mechanism for when and how cell-fate change occurs through critical transitions.

Multiomic spatial landscape of innate immune cells at human central nervous system borders

The innate immune compartment of the human central nervous system (CNS) is highly diverse and includes several immune-cell populations such as macrophages that are frequent in the brain parenchyma (microglia) and less numerous at the brain interfaces as CNS-associated macrophages (CAMs). Due to their scantiness and particular location, little is known about the presence of temporally and spatially restricted CAM subclasses during development, health and perturbation. Here we combined single-cell RNA sequencing, time-of-flight mass cytometry and single-cell spatial transcriptomics with fate mapping and advanced immunohistochemistry to comprehensively characterize the immune system at human CNS interfaces with over 356,000 analyzed transcriptomes from 102 individuals. We also provide a comprehensive analysis of resident and engrafted myeloid cells in the brains of 15 individuals with peripheral blood stem cell transplantation, revealing compartment-specific engraftment rates across different CNS interfaces. Integrated multiomic and high-resolution spatial transcriptome analysis of anatomically dissected glioblastoma samples shows regionally distinct myeloid cell-type distributions driven by hypoxia. Notably, the glioblastoma-associated hypoxia response was distinct from the physiological hypoxia response in fetal microglia and CAMs. Our results highlight myeloid diversity at the interfaces of the human CNS with the periphery and provide insights into the complexities of the human brain's immune system.

Simple Tool for Rapidly Assessing the Quality of Multiplexed Single Cell Proteomics Data

Recent advances in the sensitivity and speed of mass spectrometers coupled with improved sample preparation methods have enabled the field of single cell proteomics to proliferate. While heavy development is occurring in the label free space, dramatic improvements in throughput are provided by multiplexing with tandem mass tags. Hundreds or thousands of single cells can be analyzed with this method, yielding large data sets which may contain poor data arising from loss of material during cell sorting or poor digestion, labeling, and lysis. To date, no tools have been described that can assess data quality prior to data processing. We present herein a lightweight python script and accompanying graphic user interface that can rapidly quantify reporter ion peaks within each MS/MS spectrum in a file. With simple summary reports, we can identify single cell samples that fail to pass a set quality threshold, thus reducing analysis time waste. In addition, this tool, Diagnostic Ion Data Analysis Reduction (DIDAR), will create reduced MGF files containing only spectra possessing a user-specified number of single cell reporter ions. By reducing the number of spectra that have excessive zero values, we can speed up sample processing with little loss in data completeness as these spectra are removed in later stages in data processing workflows. DIDAR and the DIDAR GUI are compatible with all modern operating systems and are available at: https://github.com/orsburn/DIDARSCPQC. All files described in this study are available at www.massive.ucsd.edu as accession MSV000088887.

Imputation for Single-cell RNA-seq Data with Non-negative Matrix Factorization and Transfer Learning

Single-cell RNA sequencing (scRNA-seq) has been proven to be an effective technology for investigating the heterogeneity and transcriptome dynamics due to the single-cell resolution. However, one of the major problems for data obtained by scRNA-seq is excessive zeros in the count matrix, which hinders the downstream analysis enormously. Here, we present a method that integrates non-negative matrix factorization and transfer learning (NMFTL) to impute the scRNA-seq data. It borrows gene expression information from the additional dataset and adds graph-regularized terms to the decomposed matrices. These strategies not only maintain the intrinsic geometrical structure of the data itself but also further improve the accuracy of estimating the expression values by adding the transfer term in the model. The real data analysis result demonstrates that the proposed method outperforms the existing matrix-factorization-based imputation methods in recovering dropout entries, preserving gene-to-gene and cell-to-cell relationships, and in the downstream analysis, such as cell clustering analysis, the proposed method also has a good performance. For convenience, we have implemented the "NMFTL" method with R scripts, which could be available at https://github.com/FocusPaka/NMFTL.

Identification of cytokine-induced cell communications by pan-cancer meta-analysis

Cancer immune responses are complex cellular processes in which cytokine-receptor interactions play central roles in cancer development and response to therapy; dysregulated cytokine-receptor communication may lead to pathological processes, including cancer, autoimmune diseases, and cytokine storm; however, our knowledge regarding cytokine-mediated cell-cell communication (CCI) in different cancers remains limited. The present study presents a single-cell and pan-cancer-level transcriptomics integration of 41,900 cells across 25 cancer types. We developed a single-cell method to actively express 62 cytokine-receptor pairs to reveal stable cytokine-mediated cell communications involving 84 cytokines and receptors. The correlation between the sample-based CCI profile and the interactome analysis indicates multiple cytokine-receptor modules including TGFB1, IL16ST, IL15, and the PDGF family. Some isolated cytokine interactions, such as FN1-IL17RC, displayed diverse functions within over ten single-cell transcriptomics datasets. Further functional enrichment analysis revealed that the constructed cytokine-receptor interaction map is associated with the positive regulation of multiple immune response pathways. Using public TCGA pan-cancer mutational data, co-mutational analysis of the cytokines and receptors provided significant co-occurrence features, implying the existence of cooperative mechanisms. Analysis of 10,967 samples from 32 TCGA cancer types revealed that the 84 cytokine and receptor genes are significantly associated with clinical survival time. Interestingly, the tumor samples with mutations in any of the 84 cytokines and receptors have a substantially higher mutational burden, offering insights into antitumor immune regulation and response. Clinical cancer stage information revealed that tumor samples with mutations in any of the 84 cytokines and receptors stratify into earlier tumor stages, with unique cellular compositions and clinical outcomes. This study provides a comprehensive cytokine-receptor atlas of the cellular architecture in multiple cancers at the single-cell level.

TsImpute: an accurate two-step imputation method for single-cell RNA-seq data

MOTIVATION

Single-cell RNA sequencing (scRNA-seq) technology has enabled discovering gene expression patterns at single cell resolution. However, due to technical limitations, there are usually excessive zeros, called "dropouts," in scRNA-seq data, which may mislead the downstream analysis. Therefore, it is crucial to impute these dropouts to recover the biological information.

RESULTS

We propose a two-step imputation method called tsImpute to impute scRNA-seq data. At the first step, tsImpute adopts zero-inflated negative binomial distribution to discriminate dropouts from true zeros and performs initial imputation by calculating the expected expression level. At the second step, it conducts clustering with this modified expression matrix, based on which the final distance weighted imputation is performed. Numerical results based on both simulated and real data show that tsImpute achieves favorable performance in terms of gene expression recovery, cell clustering, and differential expression analysis.

AVAILABILITY AND IMPLEMENTATION

The R package of tsImpute is available at https://github.com/ZhengWeihuaYNU/tsImpute.

LMD: Multiscale Marker Identification in Single-cell RNA-seq Data

Accurate cell marker identification in single-cell RNA-seq data is crucial for understanding cellular diversity and function. An ideal marker is highly specific in identifying cells that are similar in terms of function and state. Current marker identification methods, commonly based on clustering and differential expression, capture general cell-type markers but often miss markers for subtypes or functional cell subsets, with their performance largely dependent on clustering quality. Moreover, cluster-independent approaches tend to favor genes that lack the specificity required to characterize regions within the transcriptomic space at multiple scales. Here we introduce Localized Marker Detector (LMD), a novel tool to identify "localized genes" - genes with expression profiles specific to certain groups of highly similar cells - thereby characterizing cellular diversity in a multi-resolution and fine-grained manner. LMD's strategy involves building a cell-cell affinity graph, diffusing the gene expression value across the cell graph, and assigning a score to each gene based on its diffusion dynamics. We show that LMD exhibits superior accuracy in recovering known cell-type markers in the Tabula Muris bone marrow dataset relative to other methods for marker identification. Notably, markers favored by LMD exhibit localized expression, whereas markers prioritized by other clustering-free algorithms are often dispersed in the transcriptomic space. We further group the markers suggested by LMD into functional gene modules to improve the separation of cell types and subtypes in a more fine-grained manner. These modules also identify other sources of variation, such as cell cycle status. In conclusion, LMD is a novel algorithm that can identify fine-grained markers for cell subtypes or functional states without relying on clustering or differential expression analysis. LMD exploits the complex interactions among cells and reveals cellular diversity at high resolution.

A novel f-divergence based generative adversarial imputation method for scRNA-seq data analysis

Comprehensive analysis of single-cell RNA sequencing (scRNA-seq) data can enhance our understanding of cellular diversity and aid in the development of personalized therapies for individuals. The abundance of missing values, known as dropouts, makes the analysis of scRNA-seq data a challenging task. Most traditional methods made assumptions about specific distributions for missing values, which limit their capability to capture the intricacy of high-dimensional scRNA-seq data. Moreover, the imputation performance of traditional methods decreases with higher missing rates. We propose a novel f-divergence based generative adversarial imputation method, called sc-fGAIN, for the scRNA-seq data imputation. Our studies identify four f-divergence functions, namely cross-entropy, Kullback-Leibler (KL), reverse KL, and Jensen-Shannon, that can be effectively integrated with the generative adversarial imputation network to generate imputed values without any assumptions, and mathematically prove that the distribution of imputed data using sc-fGAIN algorithm is same as the distribution of original data. Real scRNA-seq data analysis has shown that, compared to many traditional methods, the imputed values generated by sc-fGAIN algorithm have a smaller root-mean-square error, and it is robust to varying missing rates, moreover, it can reduce imputation variability. The flexibility offered by the f-divergence allows the sc-fGAIN method to accommodate various types of data, making it a more universal approach for imputing missing values of scRNA-seq data.

Evaluating Proteomics Imputation Methods with Improved Criteria

Quantitative measurements produced by tandem mass spectrometry proteomics experiments typically contain a large proportion of missing values. Missing values hinder reproducibility, reduce statistical power, and make it difficult to compare across samples or experiments. Although many methods exist for imputing missing values, in practice, the most commonly used methods are among the worst performing. Furthermore, previous benchmarking studies have focused on relatively simple measurements of error such as the mean-squared error between imputed and held-out values. Here we evaluate the performance of commonly used imputation methods using three practical, "downstream-centric" criteria. These criteria measure the ability to identify differentially expressed peptides, generate new quantitative peptides, and improve the peptide lower limit of quantification. Our evaluation comprises several experiment types and acquisition strategies, including data-dependent and data-independent acquisition. We find that imputation does not necessarily improve the ability to identify differentially expressed peptides but that it can identify new quantitative peptides and improve the peptide lower limit of quantification. We find that MissForest is generally the best performing method per our downstream-centric criteria. We also argue that existing imputation methods do not properly account for the variance of peptide quantifications and highlight the need for methods that do.

Psychiatric risk gene Transcription Factor 4 (TCF4) regulates the density and connectivity of distinct inhibitory interneuron subtypes

Transcription factor 4 (TCF4) is a basic helix-loop-helix transcription factor that is implicated in a variety of psychiatric disorders including autism spectrum disorder (ASD), major depression, and schizophrenia. Autosomal dominant mutations in TCF4 are causal for a specific ASD called Pitt-Hopkins Syndrome (PTHS). However, our understanding of etiological and pathophysiological mechanisms downstream of TCF4 mutations is incomplete. Single cell sequencing indicates TCF4 is highly expressed in GABAergic interneurons (INs). Here, we performed cell-type specific expression analysis (CSEA) and cellular deconvolution (CD) on bulk RNA sequencing data from 5 different PTHS mouse models. Using CSEA we observed differentially expressed genes (DEGs) were enriched in parvalbumin expressing (PV+) INs and CD predicted a reduction in the PV+ INs population. Therefore, we investigated the role of TCF4 in regulating the development and function of INs in the Tcf4+/tr mouse model of PTHS. In Tcf4+/tr mice, immunohistochemical (IHC) analysis of subtype-specific IN markers and reporter mice identified reductions in PV+, vasoactive intestinal peptide (VIP+), and cortistatin (CST+) expressing INs in the cortex and cholinergic (ChAT+) INs in the striatum, with the somatostatin (SST+) IN population being spared. The reduction of these specific IN populations led to cell-type specific alterations in the balance of excitatory and inhibitory inputs onto PV+ and VIP+ INs and excitatory pyramidal neurons within the cortex. These data indicate TCF4 is a critical regulator of the development of specific subsets of INs and highlight the inhibitory network as an important source of pathophysiology in PTHS.

Deep Learning Methods for Omics Data Imputation

One common problem in omics data analysis is missing values, which can arise due to various reasons, such as poor tissue quality and insufficient sample volumes. Instead of discarding missing values and related data, imputation approaches offer an alternative means of handling missing data. However, the imputation of missing omics data is a non-trivial task. Difficulties mainly come from high dimensionality, non-linear or non-monotonic relationships within features, technical variations introduced by sampling methods, sample heterogeneity, and the non-random missingness mechanism. Several advanced imputation methods, including deep learning-based methods, have been proposed to address these challenges. Due to its capability of modeling complex patterns and relationships in large and high-dimensional datasets, many researchers have adopted deep learning models to impute missing omics data. This review provides a comprehensive overview of the currently available deep learning-based methods for omics imputation from the perspective of deep generative model architectures such as autoencoder, variational autoencoder, generative adversarial networks, and Transformer, with an emphasis on multi-omics data imputation. In addition, this review also discusses the opportunities that deep learning brings and the challenges that it might face in this field.

A subpopulation of lipogenic brown adipocytes drives thermogenic memory

Sustained responses to transient environmental stimuli are important for survival. The mechanisms underlying long-term adaptations to temporary shifts in abiotic factors remain incompletely understood. Here, we find that transient cold exposure leads to sustained transcriptional and metabolic adaptations in brown adipose tissue, which improve thermogenic responses to secondary cold encounter. Primary thermogenic challenge triggers the delayed induction of a lipid biosynthesis programme even after cessation of the original stimulus, which protects from subsequent exposures. Single-nucleus RNA sequencing and spatial transcriptomics reveal that this response is driven by a lipogenic subpopulation of brown adipocytes localized along the perimeter of Ucp1hi adipocytes. This lipogenic programme is associated with the production of acylcarnitines, and supplementation of acylcarnitines is sufficient to recapitulate improved secondary cold responses. Overall, our data highlight the importance of heterogenous brown adipocyte populations for 'thermogenic memory', which may have therapeutic implications for leveraging short-term thermogenesis to counteract obesity.

Unique functional responses differentially map onto genetic subtypes of dopamine neurons

Dopamine neurons are characterized by their response to unexpected rewards, but they also fire during movement and aversive stimuli. Dopamine neuron diversity has been observed based on molecular expression profiles; however, whether different functions map onto such genetic subtypes remains unclear. In this study, we established that three genetic dopamine neuron subtypes within the substantia nigra pars compacta, characterized by the expression of Slc17a6 (Vglut2), Calb1 and Anxa1, each have a unique set of responses to rewards, aversive stimuli and accelerations and decelerations, and these signaling patterns are highly correlated between somas and axons within subtypes. Remarkably, reward responses were almost entirely absent in the Anxa1+ subtype, which instead displayed acceleration-correlated signaling. Our findings establish a connection between functional and genetic dopamine neuron subtypes and demonstrate that molecular expression patterns can serve as a common framework to dissect dopaminergic functions.

Decoding Human Biology and Disease Using Single-cell Omics Technologies

Over the past decade, advances in single-cell omics (SCO) technologies have enabled the investigation of cellular heterogeneity at an unprecedented resolution and scale, opening a new avenue for understanding human biology and disease. In this review, we summarize the developments of sequencing-based SCO technologies and computational methods, and focus on considerable insights acquired from SCO sequencing studies to understand normal and diseased properties, with a particular emphasis on cancer research. We also discuss the technological improvements of SCO and its possible contribution to fundamental research of the human, as well as its great potential in clinical diagnoses and personalized therapies of human disease.

Evaluation of zero counts to better understand the discrepancies between bulk and single-cell RNA-Seq platforms

Recent advances in sample preparation and sequencing technology have made it possible to profile the transcriptomes of individual cells using single-cell RNA sequencing (scRNA-Seq). Compared to bulk RNA-Seq data, single-cell data often contain a higher percentage of zero reads, mainly due to lower sequencing depth per cell, which affects mostly measurements of low-expression genes. However, discrepancies between platforms are observed regardless of expression level. Using four paired datasets with multiple samples each, we investigated technical and biological factors that can contribute to this expression shift. Using two separate machine learning models we found that, in addition to expression level, RNA integrity, gene or UTR3 length, and the number of transcripts potentially also influence the occurrence of zeros. These findings could enable the development of novel analytical methods for cross-platform expression shift correction. We also identified genes and biological pathways in our diverse datasets that consistently showed differences when assessed at the single cell versus bulk level to assist in interpreting analysis across transcriptomic platforms. At the gene level, 25 genes (0.12%) were found in all datasets as discordant, but at the pathway level, 7 pathways (2.02%) showed shared enrichment in discordant genes.

DISCERN: deep single-cell expression reconstruction for improved cell clustering and cell subtype and state detection

BACKGROUND

Single-cell sequencing provides detailed insights into biological processes including cell differentiation and identity. While providing deep cell-specific information, the method suffers from technical constraints, most notably a limited number of expressed genes per cell, which leads to suboptimal clustering and cell type identification.

RESULTS

Here, we present DISCERN, a novel deep generative network that precisely reconstructs missing single-cell gene expression using a reference dataset. DISCERN outperforms competing algorithms in expression inference resulting in greatly improved cell clustering, cell type and activity detection, and insights into the cellular regulation of disease. We show that DISCERN is robust against differences between batches and is able to keep biological differences between batches, which is a common problem for imputation and batch correction algorithms. We use DISCERN to detect two unseen COVID-19-associated T cell types, cytotoxic CD4+ and CD8+ Tc2 T helper cells, with a potential role in adverse disease outcome. We utilize T cell fraction information of patient blood to classify mild or severe COVID-19 with an AUROC of 80% that can serve as a biomarker of disease stage. DISCERN can be easily integrated into existing single-cell sequencing workflow.

CONCLUSIONS

Thus, DISCERN is a flexible tool for reconstructing missing single-cell gene expression using a reference dataset and can easily be applied to a variety of data sets yielding novel insights, e.g., into disease mechanisms.

Elimination of oncogenic KRAS in genetic mouse models eradicates pancreatic cancer by inducing FAS-dependent apoptosis by CD8+ T cells

Oncogenic KRASG12D (KRAS∗) is critical for the initiation and maintenance of pancreatic ductal adenocarcinoma (PDAC) and is a known repressor of tumor immunity. Conditional elimination of KRAS∗ in genetic mouse models of PDAC leads to the reactivation of FAS, CD8+ T cell-mediated apoptosis, and complete eradication of tumors. KRAS∗ elimination recruits activated CD4+ and CD8+ T cells and promotes the activation of antigen-presenting cells. Mechanistically, KRAS∗-mediated immune evasion involves the epigenetic regulation of Fas death receptor in cancer cells, via methylation of its promoter region. Furthermore, analysis of human RNA sequencing identifies that high KRAS expression in PDAC tumors shows a lower proportion of CD8+ T cells and demonstrates shorter survival compared with tumors with low KRAS expression. This study highlights the role of CD8+ T cells in the eradication of PDAC following KRAS∗ elimination and provides a rationale for the combination of KRAS∗ targeting with immunotherapy to control PDAC.

A super-enhancer-regulated RNA-binding protein cascade drives pancreatic cancer

Pancreatic ductal adenocarcinoma (PDAC) is a lethal malignancy in need of new therapeutic options. Using unbiased analyses of super-enhancers (SEs) as sentinels of core genes involved in cell-specific function, here we uncover a druggable SE-mediated RNA-binding protein (RBP) cascade that supports PDAC growth through enhanced mRNA translation. This cascade is driven by a SE associated with the RBP heterogeneous nuclear ribonucleoprotein F, which stabilizes protein arginine methyltransferase 1 (PRMT1) to, in turn, control the translational mediator ubiquitin-associated protein 2-like. All three of these genes and the regulatory SE are essential for PDAC growth and coordinately regulated by the Myc oncogene. In line with this, modulation of the RBP network by PRMT1 inhibition reveals a unique vulnerability in Myc-high PDAC patient organoids and markedly reduces tumor growth in male mice. Our study highlights a functional link between epigenetic regulation and mRNA translation and identifies components that comprise unexpected therapeutic targets for PDAC.

TimeTalk uses single-cell RNA-seq datasets to decipher cell-cell communication during early embryo development

Early embryonic development is a dynamic process that relies on proper cell-cell communication to form a correctly patterned embryo. Early embryo development-related ligand-receptor pairs (eLRs) have been shown to guide cell fate decisions and morphogenesis. However, the scope of eLRs and their influence on early embryo development remain elusive. Here, we developed a computational framework named TimeTalk from integrated public time-course mouse scRNA-seq datasets to decipher the secret of eLRs. Extensive validations and analyses were performed to ensure the involvement of identified eLRs in early embryo development. Process analysis identified that eLRs could be divided into six temporal windows corresponding to sequential events in the early embryo development process. With the interpolation strategy, TimeTalk is powerful in revealing paracrine settings and studying cell-cell communication during early embryo development. Furthermore, by using TimeTalk in the blastocyst and blastoid models, we found that the blastoid models share the core communication pathways with the epiblast and primitive endoderm lineages in the blastocysts. This result suggests that TimeTalk has transferability to other bio-dynamic processes. We also curated eLRs recognized by TimeTalk, which may provide valuable clues for understanding early embryo development and relevant disorders.

CD38-Specific CAR Integrated into CD38 Locus Driven by Different Promoters Causes Distinct Antitumor Activities of T and NK Cells

The robust and stable expression of CD38 in T-cell acute lymphoblastic leukemia (T-ALL) blasts makes CD38 chimeric antigen receptor (CAR)-T/natural killer (NK) a potential therapy for T-ALL. However, CD38 expression in normal T/NK cells causes fratricide of CD38 CAR-T/NK cells. Here a "2-in-1" gene editing strategy is developed to generate fratricide-resistant locus-specific CAR-T/NK cells. CD38-specific CAR is integrated into the disrupted CD38 locus by CRISPR/Cas9, and CAR is placed under the control of either endogenous CD38 promoter (CD38KO/KI ) or exogenous EF1α promoter (CD38KO/KI EF1α). CD38 knockout reduces fratricide and allows the expansion of CAR-T cells. Meanwhile, CD38KO/KI EF1α results in higher CAR expression than CD38KO/KI in both CAR-T and CAR-NK cells. In a mouse T-ALL model, CD38KO/KI EF1α CAR-T cells eradicate tumors better than CD38KO/KI CAR-T cells. Surprisingly, CD38KO/KI CAR-NK cells show superior tumor control than CD38KO/KI EF1α CAR-NK cells. Further investigation reveals that endogenous regulatory elements in NK cells lead to higher expression of CD38 CAR than in T cells, and the expression levels of CAR affect the therapeutic outcome of CAR-T and CAR-NK cells differently. Therefore, these results support the efficacy of CD38 CAR-T/NK against T-ALL and demonstrate that the "2-in-1" strategy can resolve fratricide and enhance tumor eradication, paving the way for clinical translation.

Humanized mouse liver reveals endothelial control of essential hepatic metabolic functions

Hepatocytes, the major metabolic hub of the body, execute functions that are human-specific, altered in human disease, and currently thought to be regulated through endocrine and cell-autonomous mechanisms. Here, we show that key metabolic functions of human hepatocytes are controlled by non-parenchymal cells (NPCs) in their microenvironment. We developed mice bearing human hepatic tissue composed of human hepatocytes and NPCs, including human immune, endothelial, and stellate cells. Humanized livers reproduce human liver architecture, perform vital human-specific metabolic/homeostatic processes, and model human pathologies, including fibrosis and non-alcoholic fatty liver disease (NAFLD). Leveraging species mismatch and lipidomics, we demonstrate that human NPCs control metabolic functions of human hepatocytes in a paracrine manner. Mechanistically, we uncover a species-specific interaction whereby WNT2 secreted by sinusoidal endothelial cells controls cholesterol uptake and bile acid conjugation in hepatocytes through receptor FZD5. These results reveal the essential microenvironmental regulation of hepatic metabolism and its human-specific aspects.

Epstein-Barr virus evades restrictive host chromatin closure by subverting B cell activation and germinal center regulatory loci

Chromatin accessibility fundamentally governs gene expression and biological response programs that can be manipulated by pathogens. Here we capture dynamic chromatin landscapes of individual B cells during Epstein-Barr virus (EBV) infection. EBV+ cells that exhibit arrest via antiviral sensing and proliferation-linked DNA damage experience global accessibility reduction. Proliferative EBV+ cells develop expression-linked architectures and motif accessibility profiles resembling in vivo germinal center (GC) phenotypes. Remarkably, EBV elicits dark zone (DZ), light zone (LZ), and post-GC B cell chromatin features despite BCL6 downregulation. Integration of single-cell assay for transposase-accessible chromatin sequencing (scATAC-seq), single-cell RNA sequencing (scRNA-seq), and chromatin immunoprecipitation sequencing (ChIP-seq) data enables genome-wide cis-regulatory predictions implicating EBV nuclear antigens (EBNAs) in phenotype-specific control of GC B cell activation, survival, and immune evasion. Knockouts validate bioinformatically identified regulators (MEF2C and NFE2L2) of EBV-induced GC phenotypes and EBNA-associated loci that regulate gene expression (CD274/PD-L1). These data and methods can inform high-resolution investigations of EBV-host interactions, B cell fates, and virus-mediated lymphomagenesis.

[Imputation method for dropout in single-cell transcriptome data]

Single-cell transcriptome sequencing (scRNA-seq) can resolve the expression characteristics of cells in tissues with single-cell precision, enabling researchers to quantify cellular heterogeneity within populations with higher resolution, revealing potentially heterogeneous cell populations and the dynamics of complex tissues. However, the presence of a large number of technical zeros in scRNA-seq data will have an impact on downstream analysis of cell clustering, differential genes, cell annotation, and pseudotime, hindering the discovery of meaningful biological signals. The main idea to solve this problem is to make use of the potential correlation between cells and genes, and to impute the technical zeros through the observed data. Based on this, this paper reviewed the basic methods of imputing technical zeros in the scRNA-seq data and discussed the advantages and disadvantages of the existing methods. Finally, recommendations and perspectives on the use and development of the method were provided.