Diffusion maps for high-dimensional single-cell analysis of differentiation data

A new genomic framework to categorize pediatric acute myeloid leukemia

Recent studies on pediatric acute myeloid leukemia (pAML) have revealed pediatric-specific driver alterations, many of which are underrepresented in the current classification schemas. To comprehensively define the genomic landscape of pAML, we systematically categorized 887 pAML into 23 mutually distinct molecular categories, including new major entities such as UBTF or BCL11B, covering 91.4% of the cohort. These molecular categories were associated with unique expression profiles and mutational patterns. For instance, molecular categories characterized by specific HOXA or HOXB expression signatures showed distinct mutation patterns of RAS pathway genes, FLT3 or WT1, suggesting shared biological mechanisms. We show that molecular categories were strongly associated with clinical outcomes using two independent cohorts, leading to the establishment of a new prognostic framework for pAML based on these updated molecular categories and minimal residual disease. Together, this comprehensive diagnostic and prognostic framework forms the basis for future classification of pAML and treatment strategies.

Integration of Pan-Cancer Single-Cell and Spatial Transcriptomics Reveals Stromal Cell Features and Therapeutic Targets in Tumor Microenvironment

Stromal cells are physiologically essential components of the tumor microenvironment (TME) that mediates tumor development and therapeutic resistance. Development of a logical and unified system for stromal cell type identification and characterization of corresponding functional properties could help design antitumor strategies that target stromal cells. Here, we performed a pan-cancer analysis of 214,972 nonimmune stromal cells using single-cell RNA sequencing from 258 patients across 16 cancer types and analyzed spatial transcriptomics from 16 patients across seven cancer types, including six patients receiving anti-PD-1 treatment. This analysis uncovered distinct features of 39 stromal subsets across cancer types, including various functional modules, spatial locations, and clinical and therapeutic relevance. Tumor-associated PGF+ endothelial tip cells with elevated epithelial-mesenchymal transition features were enriched in immune-depleted TME and associated with poor prognosis. Fibrogenic and vascular pericytes (PC) derived from FABP4+ progenitors were two distinct tumor-associated PC subpopulations that strongly interacted with PGF+ tips, resulting in excess extracellular matrix (ECM) abundance and dysfunctional vasculature. Importantly, ECM-related cancer-associated fibroblasts enriched at the tumor boundary acted as a barrier to exclude immune cells, interacted with malignant cells to promote tumor progression, and regulated exhausted CD8+ T cells via immune checkpoint ligand-receptors (e.g., LGALS9/TIM-3) to promote immune escape. In addition, an interactive web-based tool (http://www.scpanstroma.yelab.site/) was developed for accessing, visualizing, and analyzing stromal data. Taken together, this study provides a systematic view of the highly heterogeneous stromal populations across cancer types and suggests future avenues for designing therapies to overcome the tumor-promoting functions of stromal cells.

SIGNIFICANCE

Comprehensive characterization of tumor-associated nonimmune stromal cells provides a robust resource for dissecting tumor microenvironment complexity and guiding stroma-targeted therapy development across multiple human cancer types.

scSNV-seq: high-throughput phenotyping of single nucleotide variants by coupled single-cell genotyping and transcriptomics

CRISPR screens with single-cell transcriptomic readouts are a valuable tool to understand the effect of genetic perturbations including single nucleotide variants (SNVs) associated with diseases. Interpretation of these data is currently limited as genotypes cannot be accurately inferred from guide RNA identity alone. scSNV-seq overcomes this limitation by coupling single-cell genotyping and transcriptomics of the same cells enabling accurate and high-throughput screening of SNVs. Analysis of variants across the JAK1 gene with scSNV-seq demonstrates the importance of determining the precise genetic perturbation and accurately classifies clinically observed missense variants into three functional categories: benign, loss of function, and separation of function.

Single-cell and spatial profiling identify three response trajectories to pembrolizumab and radiation therapy in triple negative breast cancer

Strategies are needed to better identify patients that will benefit from immunotherapy alone or who may require additional therapies like chemotherapy or radiotherapy to overcome resistance. Here we employ single-cell transcriptomics and spatial proteomics to profile triple negative breast cancer biopsies taken at baseline, after one cycle of pembrolizumab, and after a second cycle of pembrolizumab given with radiotherapy. Non-responders lack immune infiltrate before and after therapy and exhibit minimal therapy-induced immune changes. Responding tumors form two groups that are distinguishable by a classifier prior to therapy, with one showing high major histocompatibility complex expression, evidence of tertiary lymphoid structures, and displaying anti-tumor immunity before treatment. The other responder group resembles non-responders at baseline and mounts a maximal immune response, characterized by cytotoxic T cell and antigen presenting myeloid cell interactions, only after combination therapy, which is mirrored in a murine model of triple negative breast cancer.

MENDER: fast and scalable tissue structure identification in spatial omics data

Tissue structure identification is a crucial task in spatial omics data analysis, for which increasingly complex models, such as Graph Neural Networks and Bayesian networks, are employed. However, whether increased model complexity can effectively lead to improved performance is a notable question in the field. Inspired by the consistent observation of cellular neighborhood structures across various spatial technologies, we propose Multi-range cEll coNtext DEciphereR (MENDER), for tissue structure identification. Applied on datasets of 3 brain regions and a whole-brain atlas, MENDER, with biology-driven design, offers substantial improvements over modern complex models while automatically aligning labels across slices, despite using much less running time than the second-fastest. MENDER's identification power allows the uncovering of previously overlooked spatial domains that exhibit strong associations with brain aging. MENDER's scalability makes it freely appliable on a million-level brain spatial atlas. MENDER's discriminative power enables the differentiation of breast cancer patient subtypes obscured by single-cell analysis.

Cardiac Development at a Single-Cell Resolution

Mammalian cardiac development is a complex, multistage process. Though traditional lineage tracing studies have characterized the broad trajectories of cardiac progenitors, the advent and rapid optimization of single-cell RNA sequencing methods have yielded an ever-expanding toolkit for characterizing heterogeneous cell populations in the developing heart. Importantly, they have allowed for a robust profiling of the spatiotemporal transcriptomic landscape of the human and mouse heart, revealing the diversity of cardiac cells-myocyte and non-myocyte-over the course of development. These studies have yielded insights into novel cardiac progenitor populations, chamber-specific developmental signatures, the gene regulatory networks governing cardiac development, and, thus, the etiologies of congenital heart diseases. Furthermore, single-cell RNA sequencing has allowed for the exquisite characterization of distinct cardiac populations such as the hard-to-capture cardiac conduction system and the intracardiac immune population. Therefore, single-cell profiling has also resulted in new insights into the regulation of cardiac regeneration and injury repair. Single-cell multiomics approaches combining transcriptomics, genomics, and epigenomics may uncover an even more comprehensive atlas of human cardiac biology. Single-cell analyses of the developing and adult mammalian heart offer an unprecedented look into the fundamental mechanisms of cardiac development and the complex diseases that may arise from it.

Identification of a myofibroblast differentiation program during neonatal lung development

Alveologenesis is the final stage of lung development in which the internal surface area of the lung is increased to facilitate efficient gas exchange in the mature organism. The first phase of alveologenesis involves the formation of septal ridges (secondary septae) and the second phase involves thinning of the alveolar septa. Within secondary septa, mesenchymal cells include a transient population of alveolar myofibroblasts (MyoFB) and a stable but poorly described population of lipid rich cells that have been referred to as lipofibroblasts or matrix fibroblasts (MatFB). Using a unique Fgf18CreER lineage trace mouse line, cell sorting, single cell RNA sequencing, and primary cell culture, we have identified multiple subtypes of mesenchymal cells in the neonatal lung, including an immature progenitor cell that gives rise to mature MyoFB. We also show that the endogenous and targeted ROSA26 locus serves as a sensitive reporter for MyoFB maturation. These studies identify a myofibroblast differentiation program that is distinct form other mesenchymal cells types and increases the known repertoire of mesenchymal cell types in the neonatal lung.

Interpretable trajectory inference with single-cell Linear Adaptive Negative-binomial Expression (scLANE) testing

The rapid proliferation of trajectory inference methods for single-cell RNA-seq data has allowed researchers to investigate complex biological processes by examining underlying gene expression dynamics. After estimating a latent cell ordering, statistical models are used to determine which genes exhibit changes in expression that are significantly associated with progression through the biological trajectory. While a few techniques for performing trajectory differential expression exist, most rely on the flexibility of generalized additive models in order to account for the inherent nonlinearity of changes in gene expression. As such, the results can be difficult to interpret, and biological conclusions often rest on subjective visual inspections of the most dynamic genes. To address this challenge, we propose scLANE testing, which is built around an interpretable generalized linear model and handles nonlinearity with basis splines chosen empirically for each gene. In addition, extensions to estimating equations and mixed models allow for reliable trajectory testing under complex experimental designs. After validating the accuracy of scLANE under several different simulation scenarios, we apply it to a set of diverse biological datasets and display its ability to provide novel biological information when used downstream of both pseudotime and RNA velocity estimation methods.

Data-driven selection of analysis decisions in single-cell RNA-seq trajectory inference

Single-cell RNA sequencing (scRNA-seq) experiments have become instrumental in developmental and differentiation studies, enabling the profiling of cells at a single or multiple time-points to uncover subtle variations in expression profiles reflecting underlying biological processes. Benchmarking studies have compared many of the computational methods used to reconstruct cellular dynamics, however researchers still encounter challenges in their analysis due to uncertainties in selecting the most appropriate methods and parameters. Even among universal data processing steps used by trajectory inference methods such as feature selection and dimension reduction, trajectory methods' performances are highly dataset-specific. To address these challenges, we developed Escort, a framework for evaluating a dataset's suitability for trajectory inference and quantifying trajectory properties influenced by analysis decisions. Escort navigates single-cell trajectory analysis through data-driven assessments, reducing uncertainty and much of the decision burden associated with trajectory inference. Escort is implemented in an accessible R package and R/Shiny application, providing researchers with the necessary tools to make informed decisions during trajectory analysis and enabling new insights into dynamic biological processes at single-cell resolution.

Single-cell analysis of shared signatures and transcriptional diversity during zebrafish development

During development, animals generate distinct cell populations with specific identities, functions, and morphologies. We mapped transcriptionally distinct populations across 489,686 cells from 62 stages during wild-type zebrafish embryogenesis and early larval development (3-120 h post-fertilization). Using these data, we identified the limited catalog of gene expression programs reused across multiple tissues and their cell-type-specific adaptations. We also determined the duration each transcriptional state is present during development and identify unexpected long-term cycling populations. Focused clustering and transcriptional trajectory analyses of non-skeletal muscle and endoderm identified transcriptional profiles and candidate transcriptional regulators of understudied cell types and subpopulations, including the pneumatic duct, individual intestinal smooth muscle layers, spatially distinct pericyte subpopulations, and recently discovered best4+ cells. To enable additional discoveries, we make this comprehensive transcriptional atlas of early zebrafish development available through our website, Daniocell.

LinRace: cell division history reconstruction of single cells using paired lineage barcode and gene expression data

Lineage tracing technology using CRISPR/Cas9 genome editing has enabled simultaneous readouts of gene expressions and lineage barcodes in single cells, which allows for inference of cell lineage and cell types at the whole organism level. While most state-of-the-art methods for lineage reconstruction utilize only the lineage barcode data, methods that incorporate gene expressions are emerging. Effectively incorporating the gene expression data requires a reasonable model of how gene expression data changes along generations of divisions. Here, we present LinRace (Lineage Reconstruction with asymmetric cell division model), which integrates lineage barcode and gene expression data using asymmetric cell division model and infers cell lineages and ancestral cell states using Neighbor-Joining and maximum-likelihood heuristics. On both simulated and real data, LinRace outputs more accurate cell division trees than existing methods. With inferred ancestral states, LinRace can also show how a progenitor cell generates a large population of cells with various functionalities.

Broad H3K4me3 Domain Is Associated with Spatial Coherence during Mammalian Embryonic Development

It is well known that the chromatin states play a major role in cell-fate decision and cell-identity maintenance; however, the spatial variation of chromatin states in situ remains poorly characterized. Here, by leveraging recently available spatial-CUT&Tag data, we systematically characterized the global spatial organization of the H3K4me3 profiles in a mouse embryo. Our analysis identified a subset of genes with spatially coherent H3K4me3 patterns, which together delineate the tissue boundaries. The spatially coherent genes are strongly enriched with tissue-specific transcriptional regulators. Remarkably, their corresponding genomic loci are marked by broad H3K4me3 domains, which is distinct from the typical H3K4me3 signature. Spatial transition across tissue boundaries is associated with continuous shortening of the broad H3K4me3 domains as well as expansion of H3K27me3 domains. Our analysis reveals a strong connection between the genomic and spatial variation of chromatin states, which may play an important role in embryonic development.

SEACells infers transcriptional and epigenomic cellular states from single-cell genomics data

Metacells are cell groupings derived from single-cell sequencing data that represent highly granular, distinct cell states. Here we present single-cell aggregation of cell states (SEACells), an algorithm for identifying metacells that overcome the sparsity of single-cell data while retaining heterogeneity obscured by traditional cell clustering. SEACells outperforms existing algorithms in identifying comprehensive, compact and well-separated metacells in both RNA and assay for transposase-accessible chromatin (ATAC) modalities across datasets with discrete cell types and continuous trajectories. We demonstrate the use of SEACells to improve gene-peak associations, compute ATAC gene scores and infer the activities of critical regulators during differentiation. Metacell-level analysis scales to large datasets and is particularly well suited for patient cohorts, where per-patient aggregation provides more robust units for data integration. We use our metacells to reveal expression dynamics and gradual reconfiguration of the chromatin landscape during hematopoietic differentiation and to uniquely identify CD4 T cell differentiation and activation states associated with disease onset and severity in a Coronavirus Disease 2019 (COVID-19) patient cohort.

Multimodal spatiotemporal phenotyping of human retinal organoid development

Organoids generated from human pluripotent stem cells provide experimental systems to study development and disease, but quantitative measurements across different spatial scales and molecular modalities are lacking. In this study, we generated multiplexed protein maps over a retinal organoid time course and primary adult human retinal tissue. We developed a toolkit to visualize progenitor and neuron location, the spatial arrangements of extracellular and subcellular components and global patterning in each organoid and primary tissue. In addition, we generated a single-cell transcriptome and chromatin accessibility timecourse dataset and inferred a gene regulatory network underlying organoid development. We integrated genomic data with spatially segmented nuclei into a multimodal atlas to explore organoid patterning and retinal ganglion cell (RGC) spatial neighborhoods, highlighting pathways involved in RGC cell death and showing that mosaic genetic perturbations in retinal organoids provide insight into cell fate regulation.

Single-cell RNA sequencing reveals new subtypes of lens superficial tissue in humans

Although the cell atlas of the human ocular anterior segment of the human eye was revealed by single-nucleus RNA sequencing, whether subtypes of lens stem/progenitor cells exist among epithelial cells and the molecular characteristics of cell differentiation of the human lens remain unclear. Single-cell RNA sequencing is a powerful tool to analyse the heterogeneity of tissues at the single cell level, leading to a better understanding of the processes of cell differentiation. By profiling 18,596 cells in human lens superficial tissue through single-cell sequencing, we identified two subtypes of lens epithelial cells that specifically expressed C8orf4 and ADAMTSL4 with distinct spatial localization, a new type of fibre cells located directly adjacent to the epithelium, and a subpopulation of ADAMTSL4+ cells that might be lens epithelial stem/progenitor cells. We also found two trajectories of lens epithelial cell differentiation and changes of some important genes during differentiation.

SCALA: A complete solution for multimodal analysis of single-cell Next Generation Sequencing data

Analysis and interpretation of high-throughput transcriptional and chromatin accessibility data at single-cell (sc) resolution are still open challenges in the biomedical field. The existence of countless bioinformatics tools, for the different analytical steps, increases the complexity of data interpretation and the difficulty to derive biological insights. In this article, we present SCALA, a bioinformatics tool for analysis and visualization of single-cell RNA sequencing (scRNA-seq) and Assay for Transposase-Accessible Chromatin using sequencing (scATAC-seq) datasets, enabling either independent or integrative analysis of the two modalities. SCALA combines standard types of analysis by integrating multiple software packages varying from quality control to the identification of distinct cell populations and cell states. Additional analysis options enable functional enrichment, cellular trajectory inference, ligand-receptor analysis, and regulatory network reconstruction. SCALA is fully parameterizable, presenting data in tabular format and producing publication-ready visualizations. The different available analysis modules can aid biomedical researchers in exploring, analyzing, and visualizing their data without any prior experience in coding. We demonstrate the functionality of SCALA through two use-cases related to TNF-driven arthritic mice, handling both scRNA-seq and scATAC-seq datasets. SCALA is developed in R, Shiny and JavaScript and is mainly available as a standalone version, while an online service of more limited capacity can be found at http://scala.pavlopouloslab.info or https://scala.fleming.gr.

Spatial atlas of the mouse central nervous system at molecular resolution

Spatially charting molecular cell types at single-cell resolution across the 3D volume is critical for illustrating the molecular basis of brain anatomy and functions. Single-cell RNA sequencing has profiled molecular cell types in the mouse brain1,2, but cannot capture their spatial organization. Here we used an in situ sequencing method, STARmap PLUS3,4, to profile 1,022 genes in 3D at a voxel size of 194 × 194 × 345 nm3, mapping 1.09 million high-quality cells across the adult mouse brain and spinal cord. We developed computational pipelines to segment, cluster and annotate 230 molecular cell types by single-cell gene expression and 106 molecular tissue regions by spatial niche gene expression. Joint analysis of molecular cell types and molecular tissue regions enabled a systematic molecular spatial cell-type nomenclature and identification of tissue architectures that were undefined in established brain anatomy. To create a transcriptome-wide spatial atlas, we integrated STARmap PLUS measurements with a published single-cell RNA-sequencing atlas1, imputing single-cell expression profiles of 11,844 genes. Finally, we delineated viral tropisms of a brain-wide transgene delivery tool, AAV-PHP.eB5,6. Together, this annotated dataset provides a single-cell resource that integrates the molecular spatial atlas, brain anatomy and the accessibility to genetic manipulation of the mammalian central nervous system.

Transcriptional linkage analysis with in vivo AAV-Perturb-seq

The ever-growing compendium of genetic variants associated with human pathologies demands new methods to study genotype-phenotype relationships in complex tissues in a high-throughput manner1,2. Here we introduce adeno-associated virus (AAV)-mediated direct in vivo single-cell CRISPR screening, termed AAV-Perturb-seq, a tuneable and broadly applicable method for transcriptional linkage analysis as well as high-throughput and high-resolution phenotyping of genetic perturbations in vivo. We applied AAV-Perturb-seq using gene editing and transcriptional inhibition to systematically dissect the phenotypic landscape underlying 22q11.2 deletion syndrome3,4 genes in the adult mouse brain prefrontal cortex. We identified three 22q11.2-linked genes involved in known and previously undescribed pathways orchestrating neuronal functions in vivo that explain approximately 40% of the transcriptional changes observed in a 22q11.2-deletion mouse model. Our findings suggest that the 22q11.2-deletion syndrome transcriptional phenotype found in mature neurons may in part be due to the broad dysregulation of a class of genes associated with disease susceptibility that are important for dysfunctional RNA processing and synaptic function. Our study establishes a flexible and scalable direct in vivo method to facilitate causal understanding of biological and disease mechanisms with potential applications to identify genetic interventions and therapeutic targets for treating disease.

SnapATAC2: a fast, scalable and versatile tool for analysis of single-cell omics data

Single-cell omics technologies have ushered in a new era for the study of dynamic gene regulation in complex tissues during development and disease pathogenesis. A major computational challenge in analyzing these datasets is to project the large-scale and high dimensional data into low-dimensional space while retaining the relative relationships between cells in order to decompose the cellular heterogeneity and reconstruct cell-type-specific gene regulatory programs. Conventional dimensionality reduction methods suffer from computational inefficiency, difficulty to capture the full spectrum of cellular heterogeneity, or inability to apply across diverse molecular modalities. Here, we report a fast and nonlinear dimensionality reduction algorithm that not only more accurately captures the heterogeneities of single-cell omics data, but also features runtime and memory usage that is computational efficient and linearly proportional to cell numbers. We implement this algorithm in a Python package named SnapATAC2, and demonstrate its superior performance, remarkable scalability and general adaptability using an array of single-cell omics data types, including single-cell ATAC-seq, single-cell RNA-seq, single-cell Hi-C, and single-cell multiomics datasets.

Systems level identification of a matrisome-associated macrophage polarisation state in multi-organ fibrosis

Tissue fibrosis affects multiple organs and involves a master-regulatory role of macrophages which respond to an initial inflammatory insult common in all forms of fibrosis. The recently unravelled multi-organ heterogeneity of macrophages in healthy and fibrotic human disease suggests that macrophages expressing osteopontin (SPP1) associate with lung and liver fibrosis. However, the conservation of this SPP1+ macrophage population across different tissues and its specificity to fibrotic diseases with different etiologies remain unclear. Integrating 15 single-cell RNA-sequencing datasets to profile 235,930 tissue macrophages from healthy and fibrotic heart, lung, liver, kidney, skin, and endometrium, we extended the association of SPP1+ macrophages with fibrosis to all these tissues. We also identified a subpopulation expressing matrisome-associated genes (e.g., matrix metalloproteinases and their tissue inhibitors), functionally enriched for ECM remodelling and cell metabolism, representative of a matrisome-associated macrophage (MAM) polarisation state within SPP1+ macrophages. Importantly, the MAM polarisation state follows a differentiation trajectory from SPP1+ macrophages and is associated with a core set of regulon activity. SPP1+ macrophages without the MAM polarisation state (SPP1+MAM-) show a positive association with ageing lung in mice and humans. These results suggest an advanced and conserved polarisation state of SPP1+ macrophages in fibrotic tissues resulting from prolonged inflammatory cues within each tissue microenvironment.

Transcriptomic profiling of tissue environments critical for post-embryonic patterning and morphogenesis of zebrafish skin

Pigment patterns and skin appendages are prominent features of vertebrate skin. In zebrafish, regularly patterned pigment stripes and an array of calcified scales form simultaneously in the skin during post-embryonic development. Understanding the mechanisms that regulate stripe patterning and scale morphogenesis may lead to the discovery of fundamental mechanisms that govern the development of animal form. To learn about cell types and signaling interactions that govern skin patterning and morphogenesis, we generated and analyzed single-cell transcriptomes of skin from wild-type fish as well as fish having genetic or transgenically induced defects in squamation or pigmentation. These data reveal a previously undescribed population of epidermal cells that express transcripts encoding enamel matrix proteins, suggest hormonal control of epithelial-mesenchymal signaling, clarify the signaling network that governs scale papillae development, and identify a critical role for the hypodermis in supporting pigment cell development. Additionally, these comprehensive single-cell transcriptomic data representing skin phenotypes of biomedical relevance should provide a useful resource for accelerating the discovery of mechanisms that govern skin development and homeostasis.

Placental growth factor exerts a dual function for cardiomyogenesis and vasculogenesis during heart development

Cardiogenic growth factors play important roles in heart development. Placental growth factor (PLGF) has previously been reported to have angiogenic effects; however, its potential role in cardiogenesis has not yet been determined. We analyze single-cell RNA-sequencing data derived from human and primate embryonic hearts and find PLGF shows a biphasic expression pattern, as it is expressed specifically on ISL1+ second heart field progenitors at an earlier stage and on vascular smooth muscle cells (SMCs) and endothelial cells (ECs) at later stages. Using chemically modified mRNAs (modRNAs), we generate a panel of cardiogenic growth factors and test their effects on enhancing cardiomyocyte (CM) and EC induction during different stages of human embryonic stem cell (hESC) differentiations. We discover that only the application of PLGF modRNA at early time points of hESC-CM differentiation can increase both CM and EC production. Conversely, genetic deletion of PLGF reduces generation of CMs, SMCs and ECs in vitro. We also confirm in vivo beneficial effects of PLGF modRNA for development of human heart progenitor-derived cardiac muscle grafts on murine kidney capsules. Further, we identify the previously unrecognized PLGF-related transcriptional networks driven by EOMES and SOX17. These results shed light on the dual cardiomyogenic and vasculogenic effects of PLGF during heart development.

Deep learning applications in single-cell genomics and transcriptomics data analysis

Traditional bulk sequencing methods are limited to measuring the average signal in a group of cells, potentially masking heterogeneity, and rare populations. The single-cell resolution, however, enhances our understanding of complex biological systems and diseases, such as cancer, the immune system, and chronic diseases. However, the single-cell technologies generate massive amounts of data that are often high-dimensional, sparse, and complex, thus making analysis with traditional computational approaches difficult and unfeasible. To tackle these challenges, many are turning to deep learning (DL) methods as potential alternatives to the conventional machine learning (ML) algorithms for single-cell studies. DL is a branch of ML capable of extracting high-level features from raw inputs in multiple stages. Compared to traditional ML, DL models have provided significant improvements across many domains and applications. In this work, we examine DL applications in genomics, transcriptomics, spatial transcriptomics, and multi-omics integration, and address whether DL techniques will prove to be advantageous or if the single-cell omics domain poses unique challenges. Through a systematic literature review, we have found that DL has not yet revolutionized the most pressing challenges of the single-cell omics field. However, using DL models for single-cell omics has shown promising results (in many cases outperforming the previous state-of-the-art models) in data preprocessing and downstream analysis. Although developments of DL algorithms for single-cell omics have generally been gradual, recent advances reveal that DL can offer valuable resources in fast-tracking and advancing research in single-cell.

An introduction to representation learning for single-cell data analysis

Single-cell-resolved systems biology methods, including omics- and imaging-based measurement modalities, generate a wealth of high-dimensional data characterizing the heterogeneity of cell populations. Representation learning methods are routinely used to analyze these complex, high-dimensional data by projecting them into lower-dimensional embeddings. This facilitates the interpretation and interrogation of the structures, dynamics, and regulation of cell heterogeneity. Reflecting their central role in analyzing diverse single-cell data types, a myriad of representation learning methods exist, with new approaches continually emerging. Here, we contrast general features of representation learning methods spanning statistical, manifold learning, and neural network approaches. We consider key steps involved in representation learning with single-cell data, including data pre-processing, hyperparameter optimization, downstream analysis, and biological validation. Interdependencies and contingencies linking these steps are also highlighted. This overview is intended to guide researchers in the selection, application, and optimization of representation learning strategies for current and future single-cell research applications.

Progressive plasticity during colorectal cancer metastasis

Metastasis is the principal cause of cancer death, yet we lack an understanding of metastatic cell states, their relationship to primary tumor states, and the mechanisms by which they transition. In a cohort of biospecimen trios from same-patient normal colon, primary and metastatic colorectal cancer, we show that while primary tumors largely adopt LGR5 + intestinal stem-like states, metastases display progressive plasticity. Loss of intestinal cell states is accompanied by reprogramming into a highly conserved fetal progenitor state, followed by non-canonical differentiation into divergent squamous and neuroendocrine-like states, which is exacerbated by chemotherapy and associated with poor patient survival. Using matched patient-derived organoids, we demonstrate that metastatic cancer cells exhibit greater cell-autonomous multilineage differentiation potential in response to microenvironment cues than their intestinal lineage-restricted primary tumor counterparts. We identify PROX1 as a stabilizer of intestinal lineage in the fetal progenitor state, whose downregulation licenses non-canonical reprogramming.

Infectious neutrophil deployment is regulated by resolvin D4

Neutrophils reside in the bone marrow (BM), ready for deployment to sites of injury/infection, initiating inflammation and its resolution. Here, we report that distal infections signal to the BM via resolvins to regulate granulopoiesis and BM neutrophil deployment. Emergency granulopoiesis during peritonitis evoked changes in BM resolvin D1 (RvD1) and BM RvD4. We found that leukotriene B4 stimulates neutrophil deployment. RvD1 and RvD4 each limited neutrophilic infiltration to infections, and differently regulated BM myeloid populations: RvD1 increased reparative monocytes, and RvD4 regulated granulocytes. RvD4 disengaged emergency granulopoiesis, prevented excess BM neutrophil deployment, and acted on granulocyte progenitors. RvD4 also stimulated exudate neutrophil, monocyte, and macrophage phagocytosis, and enhanced bacterial clearance. This mediator accelerated both neutrophil apoptosis and clearance by macrophages, thus expediting the resolution phase of inflammation. RvD4 stimulated phosphorylation of ERK1/2 and STAT3 in human BM-aspirate-derived granulocytes. RvD4 in the 1 to 100 nM range stimulated whole-blood neutrophil phagocytosis of Escherichia coli. RvD4 increased BM macrophage efferocytosis of neutrophils. Together, these results demonstrate the novel functions of resolvins in granulopoiesis and neutrophil deployment, contributing to the resolution of infectious inflammation.

Computational Methods for Single-Cell Proteomics

Advances in single-cell proteomics technologies have resulted in high-dimensional datasets comprising millions of cells that are capable of answering key questions about biology and disease. The advent of these technologies has prompted the development of computational tools to process and visualize the complex data. In this review, we outline the steps of single-cell and spatial proteomics analysis pipelines. In addition to describing available methods, we highlight benchmarking studies that have identified advantages and pitfalls of the currently available computational toolkits. As these technologies continue to advance, robust analysis tools should be developed in tandem to take full advantage of the potential biological insights provided by these data.

NSD2 maintains lineage plasticity and castration-resistance in neuroendocrine prostate cancer

The clinical use of potent androgen receptor (AR) inhibitors has promoted the emergence of novel subtypes of metastatic castration-resistant prostate cancer (mCRPC), including neuroendocrine prostate cancer (CRPC-NE), which is highly aggressive and lethal 1 . These mCRPC subtypes display increased lineage plasticity and often lack AR expression 2-5 . Here we show that neuroendocrine differentiation and castration-resistance in CRPC-NE are maintained by the activity of Nuclear Receptor Binding SET Domain Protein 2 (NSD2) 6 , which catalyzes histone H3 lysine 36 dimethylation (H3K36me2). We find that organoid lines established from genetically-engineered mice 7 recapitulate key features of human CRPC-NE, and can display transdifferentiation to neuroendocrine states in culture. CRPC-NE organoids express elevated levels of NSD2 and H3K36me2 marks, but relatively low levels of H3K27me3, consistent with antagonism of EZH2 activity by H3K36me2. Human CRPC-NE but not primary NEPC tumors expresses high levels of NSD2, consistent with a key role for NSD2 in lineage plasticity, and high NSD2 expression in mCRPC correlates with poor survival outcomes. Notably, CRISPR/Cas9 targeting of NSD2 or expression of a dominant-negative oncohistone H3.3K36M mutant results in loss of neuroendocrine phenotypes and restores responsiveness to the AR inhibitor enzalutamide in mouse and human CRPC-NE organoids and grafts. Our findings indicate that NSD2 inhibition can reverse lineage plasticity and castration-resistance, and provide a potential new therapeutic target for CRPC-NE.

Quantifying Cell-State Densities in Single-Cell Phenotypic Landscapes using Mellon

Cell-state density characterizes the distribution of cells along phenotypic landscapes and is crucial for unraveling the mechanisms that drive cellular differentiation, regeneration, and disease. Here, we present Mellon, a novel computational algorithm for high-resolution estimation of cell-state densities from single-cell data. We demonstrate Mellon's efficacy by dissecting the density landscape of various differentiating systems, revealing a consistent pattern of high-density regions corresponding to major cell types intertwined with low-density, rare transitory states. Utilizing hematopoietic stem cell fate specification to B-cells as a case study, we present evidence implicating enhancer priming and the activation of master regulators in the emergence of these transitory states. Mellon offers the flexibility to perform temporal interpolation of time-series data, providing a detailed view of cell-state dynamics during the inherently continuous developmental processes. Scalable and adaptable, Mellon facilitates density estimation across various single-cell data modalities, scaling linearly with the number of cells. Our work underscores the importance of cell-state density in understanding the differentiation processes, and the potential of Mellon to provide new insights into the regulatory mechanisms guiding cellular fate decisions.

Conserved stromal-immune cell circuits secure B cell homeostasis and function

B cell zone reticular cells (BRCs) form stable microenvironments that direct efficient humoral immunity with B cell priming and memory maintenance being orchestrated across lymphoid organs. However, a comprehensive understanding of systemic humoral immunity is hampered by the lack of knowledge of global BRC sustenance, function and major pathways controlling BRC-immune cell interactions. Here we dissected the BRC landscape and immune cell interactome in human and murine lymphoid organs. In addition to the major BRC subsets underpinning the follicle, including follicular dendritic cells, PI16+ RCs were present across organs and species. As well as BRC-produced niche factors, immune cell-driven BRC differentiation and activation programs governed the convergence of shared BRC subsets, overwriting tissue-specific gene signatures. Our data reveal that a canonical set of immune cell-provided cues enforce bidirectional signaling programs that sustain functional BRC niches across lymphoid organs and species, thereby securing efficient humoral immunity.

An atlas of rabbit development as a model for single-cell comparative genomics

Traditionally, the mouse has been the favoured vertebrate model for biomedical research, due to its experimental and genetic tractability. However, non-rodent embryological studies highlight that many aspects of early mouse development, such as its egg-cylinder gastrulation and method of implantation, diverge from other mammals, thus complicating inferences about human development. Like the human embryo, rabbits develop as a flat-bilaminar disc. Here we constructed a morphological and molecular atlas of rabbit development. We report transcriptional and chromatin accessibility profiles for over 180,000 single cells and high-resolution histology sections from embryos spanning gastrulation, implantation, amniogenesis and early organogenesis. Using a neighbourhood comparison pipeline, we compare the transcriptional landscape of rabbit and mouse at the scale of the entire organism. We characterize the gene regulatory programmes underlying trophoblast differentiation and identify signalling interactions involving the yolk sac mesothelium during haematopoiesis. We demonstrate how the combination of both rabbit and mouse atlases can be leveraged to extract new biological insights from sparse macaque and human data. The datasets and computational pipelines reported here set a framework for a broader cross-species approach to decipher early mammalian development, and are readily adaptable to deploy single-cell comparative genomics more broadly across biomedical research.

Phagocytosis increases an oxidative metabolic and immune suppressive signature in tumor macrophages

Phagocytosis is a key macrophage function, but how phagocytosis shapes tumor-associated macrophage (TAM) phenotypes and heterogeneity in solid tumors remains unclear. Here, we utilized both syngeneic and novel autochthonous lung tumor models in which neoplastic cells express the fluorophore tdTomato (tdTom) to identify TAMs that have phagocytosed neoplastic cells in vivo. Phagocytic tdTompos TAMs upregulated antigen presentation and anti-inflammatory proteins, but downregulated classic proinflammatory effectors compared to tdTomneg TAMs. Single-cell transcriptomic profiling identified TAM subset-specific and common gene expression changes associated with phagocytosis. We uncover a phagocytic signature that is predominated by oxidative phosphorylation (OXPHOS), ribosomal, and metabolic genes, and this signature correlates with worse clinical outcome in human lung cancer. Expression of OXPHOS proteins, mitochondrial content, and functional utilization of OXPHOS were increased in tdTompos TAMs. tdTompos tumor dendritic cells also display similar metabolic changes. Our identification of phagocytic TAMs as a distinct myeloid cell state links phagocytosis of neoplastic cells in vivo with OXPHOS and tumor-promoting phenotypes.

Temporal and spatial staging of lung alveolar regeneration is determined by the grainyhead transcription factor Tfcp2l1

Alveolar epithelial type 2 (AT2) cells harbor the facultative progenitor capacity in the lung alveolus to drive regeneration after lung injury. Using single-cell transcriptomics, software-guided segmentation of tissue damage, and in vivo mouse lineage tracing, we identified the grainyhead transcription factor cellular promoter 2-like 1 (Tfcp2l1) as a regulator of this regenerative process. Tfcp2l1 loss in adult AT2 cells inhibits self-renewal and enhances AT2-AT1 differentiation during tissue regeneration. Conversely, Tfcp2l1 blunts the proliferative response to inflammatory signaling during the early acute injury phase. Tfcp2l1 temporally regulates AT2 self-renewal and differentiation in alveolar regions undergoing active regeneration. Single-cell transcriptomics and lineage tracing reveal that Tfcp2l1 regulates cell fate dynamics across the AT2-AT1 differentiation and restricts the inflammatory program in murine AT2 cells. Organoid modeling shows that Tfcp2l1 regulation of interleukin-1 (IL-1) receptor expression controlled these cell fate dynamics. These findings highlight the critical role Tfcp2l1 plays in balancing epithelial cell self-renewal and differentiation during alveolar regeneration.

P-CSN: single-cell RNA sequencing data analysis by partial cell-specific network

Although many single-cell computational methods proposed use gene expression as input, recent studies show that replacing 'unstable' gene expression with 'stable' gene-gene associations can greatly improve the performance of downstream analysis. To obtain accurate gene-gene associations, conditional cell-specific network method (c-CSN) filters out the indirect associations of cell-specific network method (CSN) based on the conditional independence of statistics. However, when there are strong connections in networks, the c-CSN suffers from false negative problem in network construction. To overcome this problem, a new partial cell-specific network method (p-CSN) based on the partial independence of statistics is proposed in this paper, which eliminates the singularity of the c-CSN by implicitly including direct associations among estimated variables. Based on the p-CSN, single-cell network entropy (scNEntropy) is further proposed to quantify cell state. The superiorities of our method are verified on several datasets. (i) Compared with traditional gene regulatory network construction methods, the p-CSN constructs partial cell-specific networks, namely, one cell to one network. (ii) When there are strong connections in networks, the p-CSN reduces the false negative probability of the c-CSN. (iii) The input of more accurate gene-gene associations further optimizes the performance of downstream analyses. (iv) The scNEntropy effectively quantifies cell state and reconstructs cell pseudo-time.

LinRace: single cell lineage reconstruction using paired lineage barcode and gene expression data

Understanding how single cells divide and differentiate into different cell types in developed organs is one of the major tasks of developmental and stem cell biology. Recently, lineage tracing technology using CRISPR/Cas9 genome editing has enabled simultaneous readouts of gene expressions and lineage barcodes in single cells, which allows for the reconstruction of the cell division tree, and even the detection of cell types and differentiation trajectories at the whole organism level. While most state-of-the-art methods for lineage reconstruction utilize only the lineage barcode data, methods that incorporate gene expression data are emerging, aiming to improve the accuracy of lineage reconstruction. However, effectively incorporating the gene expression data requires a reasonable model on how gene expression data changes along generations of divisions. Here, we present LinRace (Lineage Reconstruction with asymmetric cell division model), a method that integrates the lineage barcode and gene expression data using the asymmetric cell division model and infers cell lineage under a framework combining Neighbor Joining and maximum-likelihood heuristics. On both simulated and real data, LinRace outputs more accurate cell division trees than existing methods. Moreover, LinRace can output the cell states (cell types) of ancestral cells, which is rarely performed with existing lineage reconstruction methods. The information on ancestral cells can be used to analyze how a progenitor cell generates a large population of cells with various functionalities. LinRace is available at: https://github.com/ZhangLabGT/LinRace.

Mathematics of neural stem cells: Linking data and processes

Adult stem cells are described as a discrete population of cells that stand at the top of a hierarchy of progressively differentiating cells. Through their unique ability to self-renew and differentiate, they regulate the number of end-differentiated cells that contribute to tissue physiology. The question of how discrete, continuous, or reversible the transitions through these hierarchies are and the precise parameters that determine the ultimate performance of stem cells in adulthood are the subject of intense research. In this review, we explain how mathematical modelling has improved the mechanistic understanding of stem cell dynamics in the adult brain. We also discuss how single-cell sequencing has influenced the understanding of cell states or cell types. Finally, we discuss how the combination of single-cell sequencing technologies and mathematical modelling provides a unique opportunity to answer some burning questions in the field of stem cell biology.

Single-Cell RNA Sequencing Reveals Atlas of Yak Testis Cells

Spermatogenesis is a complex process that involves proliferation and differentiation of diploid male germ cells into haploid flagellated sperm and requires intricate interactions between testicular somatic cells and germ cells. The cellular heterogeneity of this process presents a challenge in analyzing the different cell types at various developmental stages. Single-cell RNA sequencing (scRNA-seq) provides a useful tool for exploring cellular heterogeneity. In this study, we performed a comprehensive and unbiased single-cell transcriptomic study of spermatogenesis in sexually mature 4-year-old yak using 10× Genomics scRNA-seq. Our scRNA-seq analysis identified six somatic cell types and various germ cells, including spermatogonial stem cells, spermatogonia, early-spermatocytes, late-spermatocytes, and spermatids in yak testis. Pseudo-timing analysis showed that Leydig and myoid cells originated from common progenitor cells in yaks. Moreover, functional enrichment analysis demonstrated that the top expressed genes in yak testicular somatic cells were significantly enriched in the cAMP signaling pathway, PI3K-Akt signaling pathway, MAPK signaling pathway, and ECM receptor interactions. Throughout the spermatogenesis process, genes related to spermatogenesis, cell differentiation, DNA binding, and ATP binding were expressed. Using immunohistochemical techniques, we identified candidate marker genes for spermatogonial stem cells and Sertoli cells. Our research provides new insights into yak spermatogenesis and the development of various types of cells in the testis, and presents more reliable marker proteins for in vitro culture and identification of yak spermatogonial stem cells in the later stage.

Single-cell analysis of shared signatures and transcriptional diversity during zebrafish development

During development, animals generate distinct cell populations with specific identities, functions, and morphologies. We mapped transcriptionally distinct populations across 489,686 cells from 62 stages during wild-type zebrafish embryogenesis and early larval development (3-120 hours post-fertilization). Using these data, we identified the limited catalog of gene expression programs reused across multiple tissues and their cell-type-specific adaptations. We also determined the duration each transcriptional state is present during development and suggest new long-term cycling populations. Focused analyses of non-skeletal muscle and the endoderm identified transcriptional profiles of understudied cell types and subpopulations, including the pneumatic duct, individual intestinal smooth muscle layers, spatially distinct pericyte subpopulations, and homologs of recently discovered human best4+ enterocytes. The transcriptional regulators of these populations remain unknown, so we reconstructed gene expression trajectories to suggest candidates. To enable additional discoveries, we make this comprehensive transcriptional atlas of early zebrafish development available through our website, Daniocell.

Novel insights into Hodgkin lymphoma biology by single-cell analysis

The emergence and rapid development of single-cell technologies mark a paradigm shift in cancer research. Various technology implementations represent powerful tools to understand cellular heterogeneity, identify minor cell populations that were previously hard to detect and define, and make inferences about cell-to-cell interactions at single-cell resolution. Applied to lymphoma, recent advances in single-cell RNA sequencing have broadened opportunities to delineate previously underappreciated heterogeneity of malignant cell differentiation states and presumed cell of origin, and to describe the composition and cellular subsets in the ecosystem of the tumor microenvironment (TME). Clinical deployment of an expanding armamentarium of immunotherapy options that rely on targets and immune cell interactions in the TME emphasizes the requirement for a deeper understanding of immune biology in lymphoma. In particular, classic Hodgkin lymphoma (CHL) can serve as a study paradigm because of its unique TME, featuring infrequent tumor cells among numerous nonmalignant immune cells with significant interpatient and intrapatient variability. Synergistic to advances in single-cell sequencing, multiplexed imaging techniques have added a new dimension to describing cellular cross talk in various lymphoma entities. Here, we comprehensively review recent progress using novel single-cell technologies with an emphasis on the TME biology of CHL as an application field. The described technologies, which are applicable to peripheral blood, fresh tissues, and formalin-fixed samples, hold the promise to accelerate biomarker discovery for novel immunotherapeutic approaches and to serve as future assay platforms for biomarker-informed treatment selection, including immunotherapies.

Comparative analysis of dimension reduction methods for cytometry by time-of-flight data

While experimental and informatic techniques around single cell sequencing (scRNA-seq) are advanced, research around mass cytometry (CyTOF) data analysis has severely lagged behind. CyTOF data are notably different from scRNA-seq data in many aspects. This calls for the evaluation and development of computational methods specific for CyTOF data. Dimension reduction (DR) is one of the critical steps of single cell data analysis. Here, we benchmark the performances of 21 DR methods on 110 real and 425 synthetic CyTOF samples. We find that less well-known methods like SAUCIE, SQuaD-MDS, and scvis are the overall best performers. In particular, SAUCIE and scvis are well balanced, SQuaD-MDS excels at structure preservation, whereas UMAP has great downstream analysis performance. We also find that t-SNE (along with SQuad-MDS/t-SNE Hybrid) possesses the best local structure preservation. Nevertheless, there is a high level of complementarity between these tools, so the choice of method should depend on the underlying data structure and the analytical needs.

Single-cell chemokine receptor profiles delineate the immune contexture of tertiary lymphoid structures in head and neck squamous cell carcinoma

Tertiary lymphoid structures (TLSs) are organized aggregates of immune cells associated with favourable prognosis and response to immunotherapy in cancer, but the immune architecture of TLSs remains poorly elucidated. Here, we hypothesize that the spatial architecture of leukocytes in TLSs can be reconstructed de novo, at least partially, by cell-inherent chemokine receptors profiles. Single-cell RNA-sequencing (scRNA-seq) revealed 47 subpopulations of leukocytes in head and neck squamous cell carcinoma (HNSC). Combined with bulk RNA-seq, we observed that CXCR3, CCR7, CCR6, CXCR5, and CCR1 are TLS-associated chemokine receptors. According to the spatial reference, the cellular atlas with TLS-associated chemokine receptors in HNSC TLSs was elaborately portrayed by multiplex immunohistochemistry (mIHC). Subsequently, we explored the functions and evolutionary trajectory of cells distributed in TLSs. Our investigation presents an approach to reconstructing the immune architecture of TLSs, which would help boost the antitumor immune response by inducing neogenesis TLSs in HNSC.

Single-cell analysis of bidirectional reprogramming between early embryonic states reveals mechanisms of differential lineage plasticities

Two distinct fates, pluripotent epiblast (EPI) and primitive (extra-embryonic) endoderm (PrE), arise from common progenitor cells, the inner cell mass (ICM), in mammalian embryos. To study how these sister identities are forged, we leveraged embryonic (ES) and eXtraembryonic ENdoderm (XEN) stem cells - in vitro counterparts of the EPI and PrE. Bidirectional reprogramming between ES and XEN coupled with single-cell RNA and ATAC-seq analyses uncovered distinct rates, efficiencies and trajectories of state conversions, identifying drivers and roadblocks of reciprocal conversions. While GATA4-mediated ES-to-iXEN conversion was rapid and nearly deterministic, OCT4, KLF4 and SOX2-induced XEN-to-iPS reprogramming progressed with diminished efficiency and kinetics. The dominant PrE transcriptional program, safeguarded by Gata4, and globally elevated chromatin accessibility of EPI underscored the differential plasticities of the two states. Mapping in vitro trajectories to embryos revealed reprogramming in either direction tracked along, and toggled between, EPI and PrE in vivo states without transitioning through the ICM.

Urea cycle activation triggered by host-microbiota maladaptation driving colorectal tumorigenesis

Maladaptation of host-microbiota metabolic interplay plays a critical role in colorectal cancer initiation. Here, through a combination of single-cell transcriptomics, microbiome profiling, metabonomics, and clinical analysis on colorectal adenoma and carcinoma tissues, we demonstrate that host's urea cycle metabolism is significantly activated during colorectal tumorigenesis, accompanied by the absence of beneficial bacteria with ureolytic capacity, such as Bifidobacterium, and the overabundance of pathogenic bacteria lacking ureolytic function. Urea could enter into macrophages, inhibit the binding efficiency of p-STAT1 to SAT1 promotor region, and further skew macrophages toward a pro-tumoral phenotype characterized by the accumulation of polyamines. Treating a murine model using urea cycle inhibitors or Bifidobacterium-based supplements could mitigate urea-mediated tumorigenesis. Collectively, this study highlights the utility of urea cycle inhibitors or therapeutically manipulating microbial composition using probiotics to prevent colorectal cancer.

Single-cell transcriptomes in the heart: when every epigenome counts

The response of an organ to stimuli emerges from the actions of individual cells. Recent cardiac single-cell RNA-sequencing studies of development, injury, and reprogramming have uncovered heterogeneous populations even among previously well-defined cell types, raising questions about what level of experimental resolution corresponds to disease-relevant, tissue-level phenotypes. In this review, we explore the biological meaning behind this cellular heterogeneity by undertaking an exhaustive analysis of single-cell transcriptomics in the heart (including a comprehensive, annotated compendium of studies published to date) and evaluating new models for the cardiac function that have emerged from these studies (including discussion and schematics that depict new hypotheses in the field). We evaluate the evidence to support the biological actions of newly identified cell populations and debate questions related to the role of cell-to-cell variability in development and disease. Finally, we present emerging epigenomic approaches that, when combined with single-cell RNA-sequencing, can resolve basic mechanisms of gene regulation and variability in cell phenotype.

Identification of Antibiotic Resistance in ESKAPE Pathogens through Plasmonic Nanosensors and Machine Learning

Antibiotic-resistant ESKAPE pathogens cause nosocomial infections that lead to huge morbidity and mortality worldwide. Rapid identification of antibiotic resistance is vital for the prevention and control of nosocomial infections. However, current techniques like genotype identification and antibiotic susceptibility testing are generally time-consuming and require large-scale equipment. Herein, we develop a rapid, facile, and sensitive technique to determine the antibiotic resistance phenotype among ESKAPE pathogens through plasmonic nanosensors and machine learning. Key to this technique is the plasmonic sensor array that contains gold nanoparticles functionalized with peptides differing in hydrophobicity and surface charge. The plasmonic nanosensors can interact with pathogens to generate bacterial fingerprints that alter the surface plasmon resonance (SPR) spectra of nanoparticles. In combination with machine learning, it enables the identification of antibiotic resistance among 12 ESKAPE pathogens in less than 20 min with an overall accuracy of 89.74%. This machine-learning-based approach allows for the identification of antibiotic-resistant pathogens from patients and holds great promise as a clinical tool for biomedical diagnosis.

Plasticity in airway smooth muscle differentiation during mouse lung development

It has been proposed that smooth muscle differentiation may physically sculpt airway epithelial branches in mammalian lungs. Serum response factor (SRF) acts with its co-factor myocardin to activate the expression of contractile smooth muscle markers. In the adult, however, smooth muscle exhibits a variety of phenotypes beyond contractile, and these are independent of SRF/myocardin-induced transcription. To determine whether a similar phenotypic plasticity is exhibited during development, we deleted Srf from the mouse embryonic pulmonary mesenchyme. Srf-mutant lungs branch normally, and the mesenchyme displays mechanical properties indistinguishable from controls. scRNA-seq identified an Srf-null smooth muscle cluster, wrapping the airways of mutant lungs, which lacks contractile smooth muscle markers but retains many features of control smooth muscle. Srf-null embryonic airway smooth muscle exhibits a synthetic phenotype, compared with the contractile phenotype of mature wild-type airway smooth muscle. Our findings identify plasticity in embryonic airway smooth muscle and demonstrate that a synthetic smooth muscle layer promotes airway branching morphogenesis.

Microglia drive transient insult-induced brain injury by chemotactic recruitment of CD8+ T lymphocytes

The crosstalk between the nervous and immune systems has gained increasing attention for its emerging role in neurological diseases. Radiation-induced brain injury (RIBI) remains the most common medical complication of cranial radiotherapy, and its pathological mechanisms have yet to be elucidated. Here, using single-cell RNA and T cell receptor sequencing, we found infiltration and clonal expansion of CD8⁺ T lymphocytes in the lesioned brain tissues of RIBI patients. Furthermore, by strategies of genetic or pharmacologic interruption, we identified a chemotactic action of microglia-derived CCL2/CCL8 chemokines in mediating the infiltration of CCR2⁺/CCR5⁺ CD8⁺ T cells and tissue damage in RIBI mice. Such a chemotactic axis also participated in the progression of cerebral infarction in the mouse model of ischemic injury. Our findings therefore highlight the critical role of microglia in mediating the dysregulation of adaptive immune responses and reveal a potential therapeutic strategy for non-infectious brain diseases.

Transcriptional heterogeneity in human diabetic foot wounds

Wound repair requires the coordination of multiple cell types including immune cells and tissue resident cells to coordinate healing and return of tissue function. Diabetic foot ulceration is a type of chronic wound that impacts over 4 million patients in the US and over 7 million worldwide (Edmonds et al., 2021). Yet, the cellular and molecular mechanisms that go awry in these wounds are not fully understood. Here, by profiling chronic foot ulcers from non-diabetic (NDFUs) and diabetic (DFUs) patients using single-cell RNA sequencing, we find that DFUs display transcription changes that implicate reduced keratinocyte differentiation, altered fibroblast function and lineages, and defects in macrophage metabolism, inflammation, and ECM production compared to NDFUs. Furthermore, analysis of cellular interactions reveals major alterations in several signaling pathways that are altered in DFUs. These data provide a view of the mechanisms by which diabetes alters healing of foot ulcers and may provide therapeutic avenues for DFU treatments.

IAN: Iterated Adaptive Neighborhoods for Manifold Learning and Dimensionality Estimation

Invoking the manifold assumption in machine learning requires knowledge of the manifold's geometry and dimension, and theory dictates how many samples are required. However, in most applications, the data are limited, sampling may not be uniform, and the manifold's properties are unknown; this implies that neighborhoods must adapt to the local structure. We introduce an algorithm for inferring adaptive neighborhoods for data given by a similarity kernel. Starting with a locally conservative neighborhood (Gabriel) graph, we sparsify it iteratively according to a weighted counterpart. In each step, a linear program yields minimal neighborhoods globally, and a volumetric statistic reveals neighbor outliers likely to violate manifold geometry. We apply our adaptive neighborhoods to nonlinear dimensionality reduction, geodesic computation, and dimension estimation. A comparison against standard algorithms using, for example, k-nearest neighbors, demonstrates the usefulness of our approach.

Complex hierarchical structures in single-cell genomics data unveiled by deep hyperbolic manifold learning

With the advances in single-cell sequencing techniques, numerous analytical methods have been developed for delineating cell development. However, most are based on Euclidean space, which would distort the complex hierarchical structure of cell differentiation. Recently, methods acting on hyperbolic space have been proposed to visualize hierarchical structures in single-cell RNA-seq (scRNA-seq) data and have been proven to be superior to methods acting on Euclidean space. However, these methods have fundamental limitations and are not optimized for the highly sparse single-cell count data. To address these limitations, we propose scDHMap, a model-based deep learning approach to visualize the complex hierarchical structures of scRNA-seq data in low-dimensional hyperbolic space. The evaluations on extensive simulation and real experiments show that scDHMap outperforms existing dimensionality-reduction methods in various common analytical tasks as needed for scRNA-seq data, including revealing trajectory branches, batch correction, and denoising the count matrix with high dropout rates. In addition, we extend scDHMap to visualize single-cell ATAC-seq data.