Single-cell Gene Expression Data Research Articles

Privacy of single-cell gene expression data

Privacy of single-cell gene expression data

Cell-to-cell distance that combines gene expression and gene embeddings

The application of large-language models (LLMs) to single-cell gene-expression data has introduced a new type of data that includes a gene-embedding matrix, in addition to the experimentally obtained gene-expression matrix. This paper addresses a fundamental problem in analyzing such data: how to effectively combine the information from both matrices to better define cell-to-cell distance. We identify a computationally feasible solution that demonstrates superior ability to cluster cells of the same type across all six real datasets we tested, underscoring its advantage as a measure of cell-to-cell distance.

Single-cell and spatial detection of senescent cells using DeepScence.

Accurately identifying senescent cells is essential for studying their spatial and molecular features. We developed DeepScence, a method based on deep neural networks, to identify senescent cells in single-cell and spatial transcriptomics data. DeepScence is based on CoreScence, a senescence-associated gene set we curated that incorporates information from multiple published gene sets. We demonstrate that DeepScence can accurately identify senescent cells in single-cell gene expression data collected both in vitro and in vivo, as well as in spatial transcriptomics data generated by different platforms, substantially outperforming existing methods.

A human-specific progenitor sub-domain extends neurogenesis and increases motor neuron production.

Neurogenesis lasts ~10 times longer in developing humans compared to mice, resulting in a >1,000-fold increase in the number of neurons in the CNS. To identify molecular and cellular mechanisms contributing to this difference, we studied human and mouse motor neurogenesis using a stem cell differentiation system that recapitulates species-specific scales of development. Comparison of human and mouse single-cell gene expression data identified human-specific progenitors characterized by coexpression of NKX2-2 and OLIG2 that give rise to spinal motor neurons. Unlike classical OLIG2+ motor neuron progenitors that give rise to two motor neurons each, OLIG2+/NKX2-2+ ventral motor neuron progenitors remain cycling longer, yielding ~5 times more motor neurons that are biased toward later-born, FOXP1-expressing subtypes. Knockout of NKX2-2 converts ventral motor neuron progenitors into classical motor neuron progenitors. Such new progenitors may contribute to the increased production of human motor neurons required for the generation of larger, more complex nervous systems.

Interindividual Variation in Gut Nitrergic Neuron Density Is Regulated By GDNF Levels and ETV1

The size and function of the enteric nervous system (ENS) can vary substantially between individuals. Because ENS function is involved in the etiology of a growing number of common human diseases, understanding mechanisms that regulate ENS variation is important. We analyzed RNAseq data from 41 normal adult human colon biopsies and single-cell RNA-seq data from human and mouse developing gut. To establish cause-consequence relationship we used alleles in mice that allow levels change of the candidate effector molecule in the comparable range to human samples. We used siRNA and primary neuronal cultures to define downstream molecular events and characterized gut functional changes in mice where molecular phenotypes paralleled findings in humans. We found that glial cell line-derived neurotrophic factor (GDNF) levels in the human colon vary about 5-fold and correlate strongly with nitrergic marker expression. In mice, we defined that GDNF levels are regulated via its 3' untranslated region (3' UTR) in the gastrointestinal tract and observed similar correlation between GDNF levels and nitrergic lineage development. We identified miR-9 and miR-133 as evolutionarily conserved candidates for negative regulation of GDNF expression in the gastrointestinal tract. Functionally, an increase in inhibitory nitrergic innervation results in an increase in gastrointestinal tract transit time, stool size, and water content accompanied with modestly reduced epithelial barrier function. Mechanistically, we found that GDNF levels regulate nitrergic lineage development via induction of transcription factor ETV1, corroborated by single-cell gene expression data in human and mouse developing enteric neurons. Our results reveal how normal variation in GDNF levels influence ENS size, composition, and gut function, suggesting a mechanism for well-known interindividual variation among those parameters.

Dissecting reversible and irreversible single cell state transitions from gene regulatory networks.

Understanding cell state transitions and their governing regulatory mechanisms remains one of the fundamental questions in biology. We develop a computational method, state transition inference using cross-cell correlations (STICCC), for predicting reversible and irreversible cell state transitions at single-cell resolution by using gene expression data and a set of gene regulatory interactions. The method is inspired by the fact that the gene expression time delays between regulators and targets can be exploited to infer past and future gene expression states. From applications to both simulated and experimental single-cell gene expression data, we show that STICCC-inferred vector fields capture basins of attraction and irreversible fluxes. By connecting regulatory information with systems' dynamical behaviors, STICCC reveals how network interactions influence reversible and irreversible state transitions. Compared to existing methods that infer pseudotime and RNA velocity, STICCC provides complementary insights into the gene regulation of cell state transitions.

The indispensable role of time in autonomous development

Advances in single-cell analysis have led to a picture of development largely in agreement with Waddington’s eponymous epigenetic landscape, in which a cell’s fate is determined by its basin of attraction in a high-dimensional gene-expression space. Yet conceptual gaps remain as to how a single progenitor can simultaneously generate multiple endpoints, and why time should be required of the process at all. We propose a theoretical model based on the Hamiltonian mechanics of n-dimensional rotational motion, which resolves these paradoxes. We derive the result that systems which become different from themselves over time must initially move in a direction not towards their ultimate endpoints, and propose that this process of resolving ambiguity can be quantified (in an information-theoretic sense) by the volume subtended in gene-expression space by the trajectories taken by the system towards its endpoints. We discuss the implications of this theory for the analysis of single-cell gene-expression data in studies of development.

Construction and analysis of a lysosome-dependent cell death score-based prediction model for non-small cell lung cancer

BackgroundNon-small cell lung cancer (NSCLC) is the most common type of tumor globally and the leading cause of cancer-related deaths. Although treatment strategies such as immune checkpoint inhibitors and chemotherapy have advanced, the heterogeneity among NSCLC patients results in significant variability in treatment outcomes. Studies have shown that certain patients respond poorly to immune checkpoint inhibitors, indicating that treatment response is closely related to multiple factors. Therefore, it is necessary to develop predictive models to stratify patients based on gene expression and clinical characteristics, aiming for precision therapy.ObjectiveThis study aims to construct a stratified prognostic model for NSCLC patients based on lysosome-dependent cell death (LDCD) scoring by integrating single-cell RNA sequencing (scRNA-seq) and bulk RNA sequencing data. By analyzing the immune-related characteristics of high-risk and low-risk groups, we further explored the impact of cell death patterns on lung cancer and identified potential therapeutic targets.MethodsThis study obtained single-cell RNA sequencing data and gene expression data of NSCLC patients and normal lung tissues from the GEO and TCGA databases. We used R packages such as Seurat and CellChat for data preprocessing and analysis, and performed dimensionality reduction and visualization through Principal Component Analysis (PCA) and UMAP algorithms. LASSO regression analysis was used to construct the predictive model, followed by cross-validation and ROC curve analysis. The model’s effectiveness was validated through survival analysis and immune microenvironment analysis.ResultsThe study showed a significant increase in the proportion of monocytes in NSCLC tissues, suggesting their important role in cancer progression. Cell communication analysis indicated that macrophages, smooth muscle cells, and myeloid cells exhibit strong intercellular communication during cancer progression. Using the constructed prognostic model based on 12 LDCD-related genes, we found significant differences in overall survival and immune microenvironment between the high-risk and low-risk groups.

ScDiffusion: conditional generation of high-quality single-cell data using diffusion model.

Single-cell RNA sequencing (scRNA-seq) data are important for studying the laws of life at single-cell level. However, it is still challenging to obtain enough high-quality scRNA-seq data. To mitigate the limited availability of data, generative models have been proposed to computationally generate synthetic scRNA-seq data. Nevertheless, the data generated with current models are not very realistic yet, especially when we need to generate data with controlled conditions. In the meantime, diffusion models have shown their power in generating data with high fidelity, providing a new opportunity for scRNA-seq generation. In this study, we developed scDiffusion, a generative model combining the diffusion model and foundation model to generate high-quality scRNA-seq data with controlled conditions. We designed multiple classifiers to guide the diffusion process simultaneously, enabling scDiffusion to generate data under multiple condition combinations. We also proposed a new control strategy called Gradient Interpolation. This strategy allows the model to generate continuous trajectories of cell development from a given cell state. Experiments showed that scDiffusion could generate single-cell gene expression data closely resembling real scRNA-seq data. Also, scDiffusion can conditionally produce data on specific cell types including rare cell types. Furthermore, we could use the multiple-condition generation of scDiffusion to generate cell type that was out of the training data. Leveraging the Gradient Interpolation strategy, we generated a continuous developmental trajectory of mouse embryonic cells. These experiments demonstrate that scDiffusion is a powerful tool for augmenting the real scRNA-seq data and can provide insights into cell fate research. scDiffusion is openly available at the GitHub repository https://github.com/EperLuo/scDiffusion or Zenodo https://zenodo.org/doi/10.5281/zenodo.13268742.

WENDY: Covariance dynamics based gene regulatory network inference

Determining gene regulatory network (GRN) structure is a central problem in biology, with a variety of inference methods available for different types of data. For a widely prevalent and challenging use case, namely single-cell gene expression data measured after intervention at multiple time points with unknown joint distributions, there is only one known specifically developed method, which does not fully utilize the rich information contained in this data type. We develop an inference method for the GRN in this case, netWork infErence by covariaNce DYnamics, dubbed WENDY. The core idea of WENDY is to model the dynamics of the covariance matrix, and solve this dynamics as an optimization problem to determine the regulatory relationships. To evaluate its effectiveness, we compare WENDY with other inference methods using synthetic data and experimental data. Our results demonstrate that WENDY performs well across different data sets.

Genome-wide study of half a million individuals with major depression identifies 697 independent associations, infers causal neuronal subtypes and biological targets for novel pharmacotherapies.

In a genome-wide association study (GWAS) of 685,808 individuals with major depression (MD) and 4,364,225 controls from 29 countries and across diverse and admixed ancestries, we identify 697 independent associations at 636 genetic loci, 293 of which are novel. Using fine-mapping and functional genomic datasets, we find 308 high-confidence gene associations and enrichment of postsynaptic density and receptor clustering. Leveraging new single-cell gene expression data, we conducted a causal neural cell type enrichment analysis that implicated excitatory and inhibitory midbrain and forebrain neurons, peptidergic neurons, and medium spiny neurons in MD. Critically, our findings are enriched for the targets of antidepressants and provide potential antidepressant repurposing opportunities (e.g., pregabalin and modafinil). Polygenic scores (PGS) from European ancestries explained up to 5.7% of the variance in liability to MD in European samples and PGS trained using either European or multi-ancestry data significantly predicted case control status across all four diverse ancestries. These findings represent a major advance in our understanding of MD across global populations. We provide evidence that MD GWAS reveals known and novel biological targets that may be used to target and develop pharmacotherapies addressing the considerable unmet need for effective treatment.

ProBac-seq, a bacterial single-cell RNA sequencing methodology using droplet microfluidics and large oligonucleotide probe sets.

Methods that measure the transcriptomic state of thousands of individual cells have transformed our understanding of cellular heterogeneity in eukaryotic cells since their introduction in the past decade. While simple and accessible protocols and commercial products are now available for the processing of mammalian cells, these existing technologies are incompatible with use in bacterial samples for several fundamental reasons including the absence of polyadenylation on bacterial messenger RNA, the instability of bacterial transcripts and the incompatibility of bacterial cell morphology with existing methodologies. Recently, we developed ProBac sequencing (ProBac-seq), a method that overcomes these technical difficulties and provides high-quality single-cell gene expression data from thousands of bacterial cells by using messenger RNA-specific probes. Here we provide details for designing large oligonucleotide probe sets for an organism of choice, amplifying probe sets to produce sufficient quantities for repeated experiments, adding unique molecular indexes and poly-A tails to produce finalized probes, in situ probe hybridization and single-cell encapsulation and library preparation. This protocol, from the probe amplification to the library preparation, requires ~7 d to complete. ProBac-seq offers several advantages over other methods by capturing only the desired target sequences and avoiding nondesired transcripts, such as highly abundant ribosomal RNA, thus enriching for signal that better informs on cellular state. The use of multiple probes per gene can detect meaningful single-cell signals from cells expressing transcripts to a lesser degree or those grown in minimal media and other environmentally relevant conditions in which cells are less active. ProBac-seq is also compatible with other organisms that can be profiled by in situ hybridization techniques.

Biclustering analysis on tree-shaped time-series single cell gene expression data of Caenorhabditis elegans

BackgroundIn recent years, gene clustering analysis has become a widely used tool for studying gene functions, efficiently categorizing genes with similar expression patterns to aid in identifying gene functions. Caenorhabditis elegans is commonly used in embryonic research due to its consistent cell lineage from fertilized egg to adulthood. Biologists use 4D confocal imaging to observe gene expression dynamics at the single-cell level. However, on one hand, the observed tree-shaped time-series datasets have characteristics such as non-pairwise data points between different individuals. On the other hand, the influence of cell type heterogeneity should also be considered during clustering, aiming to obtain more biologically significant clustering results.ResultsA biclustering model is proposed for tree-shaped single-cell gene expression data of Caenorhabditis elegans. Detailedly, a tree-shaped piecewise polynomial function is first employed to fit non-pairwise gene expression time series data. Then, four factors are considered in the objective function, including Pearson correlation coefficients capturing gene correlations, p-values from the Kolmogorov-Smirnov test measuring the similarity between cells, as well as gene expression size and bicluster overlapping size. After that, Genetic Algorithm is utilized to optimize the function.ConclusionThe results on the small-scale dataset analysis validate the feasibility and effectiveness of our model and are superior to existing classical biclustering models. Besides, gene enrichment analysis is employed to assess the results on the complete real dataset analysis, confirming that the discovered biclustering results hold significant biological relevance.

ScGAL: unmask tumor clonal substructure by jointly analyzing independent single-cell copy number and scRNA-seq data

BackgroundAccurately deciphering clonal copy number substructure can provide insights into the evolutionary mechanism of cancer, and clustering single-cell copy number profiles has become an effective means to unmask intra-tumor heterogeneity (ITH). However, copy numbers inferred from single-cell DNA sequencing (scDNA-seq) data are error-prone due to technically confounding factors such as amplification bias and allele-dropout, and this makes it difficult to precisely identify the ITH.ResultsWe introduce a hybrid model called scGAL to infer clonal copy number substructure. It combines an autoencoder with a generative adversarial network to jointly analyze independent single-cell copy number profiles and gene expression data from same cell line. Under an adversarial learning framework, scGAL exploits complementary information from gene expression data to relieve the effects of noise in copy number data, and learns latent representations of scDNA-seq cells for accurate inference of the ITH. Evaluation results on three real cancer datasets suggest scGAL is able to accurately infer clonal architecture and surpasses other similar methods. In addition, assessment of scGAL on various simulated datasets demonstrates its high robustness against the changes of data size and distribution. scGAL can be accessed at: https://github.com/zhyu-lab/scgal.ConclusionsJoint analysis of independent single-cell copy number and gene expression data from a same cell line can effectively exploit complementary information from individual omics, and thus gives more refined indication of clonal copy number substructure.

Inferring gene regulatory networks from single-cell multiome data using atlas-scale external data.

Existing methods for gene regulatory network (GRN) inference rely on gene expression data alone or on lower resolution bulk data. Despite the recent integration of chromatin accessibility and RNA sequencing data, learning complex mechanisms from limited independent data points still presents a daunting challenge. Here we present LINGER (Lifelong neural network for gene regulation), a machine-learning method to infer GRNs from single-cell paired gene expression and chromatin accessibility data. LINGER incorporates atlas-scale external bulk data across diverse cellular contexts and prior knowledge of transcription factor motifs as a manifold regularization. LINGER achieves a fourfold to sevenfold relative increase in accuracy over existing methods and reveals a complex regulatory landscape of genome-wide association studies, enabling enhanced interpretation of disease-associated variants and genes. Following the GRN inference from reference single-cell multiome data, LINGER enables the estimation of transcription factor activity solely from bulk or single-cell gene expression data, leveraging the abundance of available gene expression data to identify driver regulators from case-control studies.

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.

Abstract 6348: Unveiling diverse trimeric forms of STEAP1 in human prostate cancer cells: Exploring STEAP1-STEAP2 interdependence for expression and plasma membrane localization

Abstract Six-Transmembrane Epithelial Antigen of the Prostate 1 (STEAP1) is an emerging therapeutic target that is a member of the STEAP1-4 family of metalloreductases. We recently reported the preclinical characterization of xaluritamig (AMG 509), a STEAP1 XmAb® 2+1 T-cell engager molecule that is being developed for the treatment of prostate cancer (Nolan-Stevaux et al, 2023). Xaluritamig is designed to preferentially target tumor cells that express high levels of STEAP1. However, prostate tumor cells also express high levels of STEAP2. Here, we characterize the interactions between STEAP1 and STEAP2 in prostate cancer cell lines and the impact to xaluritamig activity. Bulk mRNA as well as single-cell gene expression data (scRNAseq) reveal a strong correlation between STEAP1 and STEAP2 expression in human prostate cancer, suggesting that the STEAP proteins may form both homotrimers and heterotrimers in prostate tumors. To test this idea, we investigated the structural behavior of STEAP1/2 trimers in human prostate cancer cell lines that endogenously express STEAP1 and STEAP2. Using xaluritamig and the parental anti-STEAP1 antibody, we show that STEAP1 primarily forms heterotrimers with STEAP2 in prostate cancer cells but can also assemble as homotrimers in the absence of STEAP2. Both STEAP1/2 heterotrimers and STEAP1 homotrimers can be targeted by xaluritamig, leading to T cell-dependent cellular cytotoxicity. Similarly, we found that anti-STEAP2 antibodies detected STEAP2 homotrimers when STEAP1 expression was ablated in cells, and that both STEAP1/2 heterotrimers and STEAP2 homotrimers were targeted by a STEAP2-specific T-cell engager molecule. We also demonstrate that neither STEAP1 nor STEAP2 significantly influences the mRNA or protein expression of the other. Our data elucidates the versatility of STEAP family members in forming both homotrimers and heterotrimers on the plasma membrane of human prostate cancer cells and the potential for therapeutic targeting with STEAP1/2-directed therapies. Characterizing the capability of existing STEAP1 or STEAP2 targeting therapeutics to bind their respective targets, whether within a homotrimer or heterotrimer, is crucial for a comprehensive understanding of the therapeutic potential of these drugs. Further research will help delineate the mechanism underlying the formation of STEAP1/2 trimers in cancer cells and continue advancing our understanding of this intriguing biological relationship. Citation Format: Cong Li, Lingming Liang, Joonhee Park, Olga Pryshchep, Binnaz Staley, Peng Li, Ben Alba, Xuan Liu, Fei Li, Gregory L. Moore, Matthew J. Bernett, Umesh Muchhal, John Desjarlais, Angela Coxon, Julie M. Bailis. Unveiling diverse trimeric forms of STEAP1 in human prostate cancer cells: Exploring STEAP1-STEAP2 interdependence for expression and plasma membrane localization [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6348.

Label-aware distance mitigates temporal and spatial variability for clustering and visualization of single-cell gene expression data

Clustering and visualization are essential parts of single-cell gene expression data analysis. The Euclidean distance used in most distance-based methods is not optimal. The batch effect, i.e., the variability among samples gathered from different times, tissues, and patients, introduces large between-group distance and obscures the true identities of cells. To solve this problem, we introduce Label-Aware Distance (Lad), a metric using temporal/spatial locality of the batch effect to control for such factors. We validate Lad on simulated data as well as apply it to a mouse retina development dataset and a lung dataset. We also found the utility of our approach in understanding the progression of the Coronavirus Disease 2019 (COVID-19). Lad provides better cell embedding than state-of-the-art batch correction methods on longitudinal datasets. It can be used in distance-based clustering and visualization methods to combine the power of multiple samples to help make biological findings.

Interrogations of single-cell RNA splicing landscapes with SCASL define new cell identities with physiological relevance

RNA splicing shapes the gene regulatory programs that underlie various physiological and disease processes. Here, we present the SCASL (single-cell clustering based on alternative splicing landscapes) method for interrogating the heterogeneity of RNA splicing with single-cell RNA-seq data. SCASL resolves the issue of biased and sparse data coverage on single-cell RNA splicing and provides a new scheme for classifications of cell identities. With previously published datasets as examples, SCASL identifies new cell clusters indicating potentially precancerous and early-tumor stages in triple-negative breast cancer, illustrates cell lineages of embryonic liver development, and provides fine clusters of highly heterogeneous tumor-associated CD4 and CD8 T cells with functional and physiological relevance. Most of these findings are not readily available via conventional cell clustering based on single-cell gene expression data. Our study shows the potential of SCASL in revealing the intrinsic RNA splicing heterogeneity and generating biological insights into the dynamic and functional cell landscapes in complex tissues.

MarkerMap: nonlinear marker selection for single-cell studies

Single-cell RNA-seq data allow the quantification of cell type differences across a growing set of biological contexts. However, pinpointing a small subset of genomic features explaining this variability can be ill-defined and computationally intractable. Here we introduce MarkerMap, a generative model for selecting minimal gene sets which are maximally informative of cell type origin and enable whole transcriptome reconstruction. MarkerMap provides a scalable framework for both supervised marker selection, aimed at identifying specific cell type populations, and unsupervised marker selection, aimed at gene expression imputation and reconstruction. We benchmark MarkerMap’s competitive performance against previously published approaches on real single cell gene expression data sets. MarkerMap is available as a pip installable package, as a community resource aimed at developing explainable machine learning techniques for enhancing interpretability in single-cell studies.