SEARCH · bioRxiv Science

Results for “single cell RNA sequencing”

Search indexed bioRxiv preprints in genomics, neuroscience, cell biology and bioinformatics. Read source abstracts and check manuscript versions; preprints are not peer reviewed.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 505 records · Page 28Linked to original sources

bioRxiv2017

BraCeR: Reconstruction of B-cell receptor sequences and clonality inference from single-cell RNA-sequencing

Reconstruction of antigen receptor sequences from single-cell RNA-sequencing (scRNA-seq) data allows the linking of antigen receptor usage to the full transcriptomic identity of individual B lymphocytes, without having to perform additional targeted repertoire sequencing (Rep-seq). Here we report BraCeR (freely available at https://github.com/teichlab/bracer/), an extension of TraCeR [1], for reconstruction of paired full-length B-cell receptor sequences and inference of clonality from scRNA-seq data (Supplementary Note 1).

bioinformatics↗

bioRxiv2026

Striving towards improved full-length single-cell RNA-sequencing

Full-length single-cell RNA-sequencing (scRNA-seq) methods provide superior transcript coverage and isoform resolution compared to their 3-end counterparts, but are typically limited to short-read platforms. Here, we report efforts to improve the FLASH-seq protocol and adapt it for long-read sequencing on the Oxford Nanopore Technologies (ONT) platform (FLASH-seq-ONT). We developed two plate-barcoding strategies enabling higher multiplexing: a custom PCR-ligation approach (PCR-LIG) and ONT native barcoding (NB-ONT). To support data processing, we built FSNanoporeR, a comprehensive bioinformatics pipeline for barcode demultiplexing, chimeric read detection and splitting, UMI extraction, and transcript quantification. Both barcoding strategies produced high-quality transcriptomic data from HEK293T cells, with notable differences in read length distributions. We further demonstrated that monomeric and trimeric UMIs can be reliably detected in >82% of reads, enabling accurate molecular counting at isoform resolution. However, both multiplexing approaches exhibited also critical limitations, including high chimeric read rates and index-swapping artifacts. Our results highlight both the promise and current technical hurdles of full-length single-cell long-read sequencing, and provide a practical framework for researchers considering ONT-based scRNA-seq workflows.

genomics↗

bioRxiv2026

BenchDrop-seq: a microfluidics-free platform for benchtop single-cell long-read RNA sequencing

Single-cell long-read RNA sequencing enables direct measurement of full-length transcripts but has remained difficult to deploy at scale due to reliance on microfluidic barcoding, specialized instrumentation, and high per-cell cost. Here we present BenchDrop-seq, a benchtop platform for single-cell long-read transcriptomics that leverages particle-templated partitioning for single-cell molecular barcoding and couples this workflow to Oxford Nanopore sequencing for full-length transcript capture. By integrating established bead-based partitioning chemistry with long-read sequencing and a dedicated open-source analysis pipeline for barcode recovery, alignment, and transcript quantification, BenchDrop-seq enables isoform-resolved measurements from thousands of individual cells using standard laboratory equipment. We validate the platform in both a homogeneous cell line and a heterogeneous primary tissue, demonstrating high barcode recovery, accurate gene-level quantification, and reproducible detection of cell-type-specific transcript usage that is not readily accessible to short-read assays. Together, BenchDrop-seq establishes a practical and accessible framework for single-cell long-read RNA sequencing, lowering experimental barriers while enabling transcript-level analyses in routine single-cell experiments.

genomics↗

bioRxiv2023

Human Pancreatic α-Cell Heterogeneity and Trajectory Inference Analysis Using Integrated Single Cell- and Single Nucleus-RNA Sequencing Platforms

Prior studies have shown that pancreatic -cells can transdifferentiate into {beta}-cells, and that {beta}-cells de-differentiate and are prone to acquire an -cell phenotype in type 2 diabetes (T2D). However, the specific human -cell and {beta}-cell subtypes that are involved in -to-{beta}-cell and {beta}-to--cell transitions are unknown. Here, we have integrated single cell RNA sequencing (scRNA-seq) and single nucleus RNA-seq (snRNA-seq) of isolated human islets and human islet grafts and provide additional insight into -{beta} cell fate switching. Using this approach, we make seven novel observations. 1) There are five different GCG-expressing human -cell subclusters [1, 2, -{beta}-transition 1 (AB-Tr1), -{beta}-transition 2 (AB-Tr2), and -{beta} (AB) cluster] with different transcriptome profiles in human islets from non-diabetic donors. 2) The AB subcluster displays multihormonal gene expression, inferred mostly from snRNA-seq data suggesting identification by pre-mRNA expression. 3) The 1, 2, AB-Tr1, and AB-Tr2 subclusters are enriched in genes specific for -cell function while AB cells are enriched in genes related to pancreatic progenitor and {beta}-cell pathways; 4) Trajectory inference analysis of extracted - and {beta}-cell clusters and RNA velocity/PAGA analysis suggests a bifurcate transition potential for AB towards both - and {beta}-cells. 5) Gene commonality analysis identifies ZNF385D, TRPM3, CASR, MEG3 and HDAC9 as signature for trajectories moving towards {beta}-cells and SMOC1, PLCE1, PAPPA2, ZNF331, ALDH1A1, SLC30A8, BTG2, TM4SF4, NR4A1 and PSCK2 as signature for trajectories moving towards -cells. 6) Remarkably, in contrast to the events in vitro, the AB subcluster is not identified in vivo in human islet grafts and trajectory inference analysis suggests only unidirectional transition from -to-{beta}-cells in vivo. 7) Analysis of scRNA-seq datasets from adult human T2D donor islets reveals a clear unidirectional transition from {beta}-to--cells compatible with dedifferentiation or conversion into -cells. Collectively, these studies show that snRNA-seq and scRNA-seq can be leveraged to identify transitions in the transcriptional status among human islet endocrine cell subpopulations in vitro, in vivo, in non-diabetes and in T2D. They reveal the potential gene signatures for common trajectories involved in interconversion between - and {beta}-cells and highlight the utility and power of studying single nuclear transcriptomes of human islets in vivo. Most importantly, they illustrate the importance of studying human islets in their natural in vivo setting.

cell biology↗

bioRxiv2020

Cell type-specific weighting-factors for accurate virtual single-cell RNA-sequencing of diverse organs

While single-cell RNA-sequencing (scRNA-seq) is a popular method to analyze gene expression and cellular composition at single-cell resolution, it harbors shortcomings: The failure to account for cell-to-cell variations of transcriptome-size (i.e., the total number of transcripts per cell) and also cell dissociation/processing-induced cryptic gene expression. This is particularly a problem when analyzing highly heterogeneous solid tissues/organs, which requires cell dissociation for the analysis. As a result, there exists a discrepancy between bulk RNA-seq result and virtually reconstituted bulk RNA-seq result using its composite scRNA-seq data. To fix this problem, we propose a computationally calculated coefficient, "cell type-specific weighting-factor (cWF)". Here, we introduce a concept and a method of its computation and report cWFs for 76 cell-types across 10 solid organs. Their fidelity is validated by more accurate reconstitution and deconvolution of bulk RNA-seq data of diverse solid organs using the scRNA-seq data and the cWFs of their composite cells. Furthermore, we also show that cWFs effectively predict aging-progression, implicating their diagnostic applications and also their association with aging mechanism. Our study provides an important method to solve critical limitations of scRNA-seq analysis of complex solid tissues/organs. Furthermore, our findings suggest a diagnostic utility and biological significance of cWFs.

bioinformatics↗

bioRxiv2025

scnanoseq: an nf-core pipeline for Oxford Nanopore single-cell RNA-sequencing

Recent advancements in long-read single-cell RNA sequencing (scRNA-seq) have facilitated the quantification of full-length transcripts and isoforms at the single-cell level. Historically, long-read data would need to be complemented with short-read single-cell data in order to overcome the higher sequencing errors to correctly identify cellular barcodes and unique molecular identifiers. Improvements in Oxford Nanopore sequencing, and development of novel computational methods have removed this requirement. Though these methods now exist, the limited availability of modular and portable workflows remains a challenge. Here we present, nf-core/scnanoseq, a secondary analysis pipeline for long-read single-cell and single-nuclei RNA that delivers gene and transcript-level quantification. The scnanoseq pipeline is implemented using Nextflow and is built upon the nf-core framework, enabling portability across computational environments, scalability and reproducibility of results across pipeline runs. The nf-core/scnanoseq workflow follows best practices for analyzing single-cell and single-nuclei data, performing barcode detection and correction, genome and transcriptome read alignment, unique molecular identifier deduplication, gene and transcript quantification, and extensive quality control reporting. AVAILABILITY OF DATA AND MATERIALSnf-core/scnanoseq is available at https://github.com/nf-core/scnanoseq under the MIT License and the documentation is available at https://nf-co.re/scnanoseq. The downstream analytical code for validation analysis is available at https://github.com/U-BDS/scnanoseq_analysis and all dataset sources have been disclosed under the methods section for each respective dataset.

bioinformatics↗

bioRxiv2020

FIRM: Fast Integration of single-cell RNA-sequencing data across Multiple platforms

Single-cell RNA-sequencing (scRNA-seq) is being used extensively to measure the mRNA expression of individual cells from deconstructed tissues, organs, and even entire organisms to generate cell atlas references, leading to discoveries of novel cell types and deeper insight into biological trajectories. These massive datasets are usually collected from many samples using different scRNA-seq technology platforms, including the popular SMART-Seq2 (SS2) and 10X platforms. Inherent heterogeneities between platforms, tissues, and other batch effects makes scRNA-seq data difficult to compare and integrate, especially in large-scale cell atlas efforts; yet, accurate integration is essential for gaining deeper insights into cell biology. Through comprehensive data exploration, we found that accurate integration is often hampered by differences in cell-type compositions. Herein we describe FIRM, an algorithm that addresses this problem and achieves efficient and accurate integration of heterogeneous scRNA-seq datasets across multiple tissue types, platforms, and experimental batches. We applied FIRM to numerous large-scale scRNA-seq datasets from mouse, mouse lemur, and human, comparing its performance in dataset integration with other state-of-the-art methods. FIRM-integrated datasets show accurate mixing of shared cell type identities and superior preservation of original structure without overcorrection, generating robust integrated datasets for downstream exploration and analysis. It is also a facile way to transfer cell type labels and annotations from one dataset to another, making it a reliable and versatile tool for scRNA-seq analysis, especially for cell atlas data integration.

bioinformatics↗

bioRxiv2024

SCEMENT: Scalable and Memory Efficient Integration of Large-scale Single Cell RNA-sequencing Data

MotivationIntegrative analysis of large-scale single cell data collected from diverse cell populations promises an improved understanding of complex biological systems. While several algorithms have been developed for single cell RNA-sequencing data integration, many lack scalability to handle large numbers of datasets and/or millions of cells due to their memory and run time requirements. The few tools which can handle large data do so by reducing the computational burden through strategies such as subsampling of the data or selecting a reference dataset, to improve computational efficiency and scalability. Such shortcuts however hamper accuracy of downstream analyses, especially those requiring quantitative gene expression information. ResultsWe present SCEMENT, a SCalablE and Memory-Efficient iNTegration method to overcome these limitations. Our new parallel algorithm builds upon and extends the linear regression model previously applied in ComBat, to an unsupervised sparse matrix setting to enable accurate integration of diverse and large collections of single cell RNA-sequencing data. Using tens to hundreds of real single cell RNA-seq datasets, we show that SCEMENT outperforms ComBat as well as FastIntegration and Scanorama in runtime (upto 214X faster) and memory usage (upto 17.5X less). It not only performs batch correction and integration of millions of cells in under 25 minutes, but also facilitates discovery of new rare cell-types and more robust reconstruction of gene regulatory networks with full quantitative gene expression information. Availability and implementationSource code freely available for download at https://github.com/AluruLab/scement, implemented in C++ and supported on Linux. Contactaluru@cc.gatech.edu Supplementary informationSupplementary data are available at Bioinformatics online.

bioinformatics↗

bioRxiv2019

Single-cell RNA-sequencing reveals profibrotic roles of distinct epithelial and mesenchymal lineages in pulmonary fibrosis

Pulmonary fibrosis is a form of chronic lung disease characterized by pathologic epithelial remodeling and accumulation of extracellular matrix. In order to comprehensively define the cell types, mechanisms and mediators driving fibrotic remodeling in lungs with pulmonary fibrosis, we performed single-cell RNA-sequencing of single-cell suspensions from 10 non-fibrotic control and 20 PF lungs. Analysis of 114,396 cells identified 31 distinct cell types. We report a remarkable shift in epithelial cell phenotypes occurs in the peripheral lung in PF, and identify several previously unrecognized epithelial cell phenotypes including a KRT5-/KRT17+, pathologic ECM-producing epithelial cell population that was highly enriched in PF lungs. Multiple fibroblast subtypes were observed to contribute to ECM expansion in a spatially-discrete manner. Together these data provide high-resolution insights into the complexity and plasticity of the distal lung epithelium in human disease, and indicate a diversity of epithelial and mesenchymal cells contribute to pathologic lung fibrosis.\n\nOne Sentence SummarySingle-cell RNA-sequencing provides new insights into pathologic epithelial and mesenchymal remodeling in the human lung.

genomics↗

bioRxiv2023

Multimodal profiling of peripheral blood identifies proliferating circulating effector CD4+ T cells as predictors for response to integrin α4β7-blocking therapy in patients with inflammatory bowel disease

Despite the success of biological therapies in inflammatory bowel disease (IBD), patient management remains challenging due to a lack of therapy response predictors. Here we prospectively sampled two cohorts of IBD patient cohorts receiving the anti-integrin 4{beta}7 antibody vedolizumab. Samples were subjected to mass cytometry, single-cell RNA sequencing, single-cell V(D)J sequencing, serum proteomics, and multidimensional flow cytometry to comprehensively assess vedolizumab-induced immunological changes in the peripheral blood and their potential associations with treatment response. Vedolizumab induced changes in the abundance of both circulating innate and adaptive immune cell compartments and modified the T cell receptor diversity of circulating gut-homing CD4+ memory T cells. Through integration of multimodal parameters and machine learning, we identify that pretreatment activated proliferating CD4+ memory T cell abundance is associated with treatment failure, independent of clinical variables, thereby providing a reliable predictive classifier with significant implications for the personalized management of IBD patients.

immunology↗

bioRxiv2024

Cellular cartography reveals mouse prostate organization and determinants of castration resistance

Inadequate response to androgen deprivation therapy (ADT) frequently arises in prostate cancer, driven by cellular mechanisms that remain poorly understood. Here, we integrated single-cell RNA sequencing, single-cell multiomics, and spatial transcriptomics to define the transcriptional, epigenetic, and spatial basis of cell identity and castration response in the mouse prostate. Leveraging these data along with a meta-analysis of human prostates and prostate cancer, we identified cellular orthologs and key determinants of ADT response and resistance. Our findings reveal that mouse prostates harbor lobe-specific luminal epithelial cell types distinguished by unique gene regulatory modules and anatomically defined androgen-responsive transcriptional programs, indicative of divergent developmental origins. Androgen-insensitive, stem-like epithelial populations - resembling human club and hillock cells - are notably enriched in the urethra and ventral prostate but are rare in other lobes. Within the ventral prostate, we also uncovered two additional androgen-responsive luminal epithelial cell types, marked by Pbsn or Spink1 expression, which align with human luminal subsets and may define the origin of distinct prostate cancer subtypes. Castration profoundly reshaped luminal epithelial transcriptomes, with castration-resistant luminal epithelial cells activating stress-responsive and stemness programs. These transcriptional signatures are enriched in tumor cells from ADT-treated and castration-resistant prostate cancer patients, underscoring their likely role in driving treatment resistance. Collectively, our comprehensive cellular atlas of the mouse prostate illuminates the importance of lobe-specific contexts for prostate cancer modeling and reveals potential therapeutic targets to counter castration resistance. Significance StatementAndrogen deprivation therapy is a mainstay in prostate cancer treatment, yet many patients eventually develop castration-resistant disease--a lethal progression driven by poorly understood cellular mechanisms. Our study provides a comprehensive cellular map of the prostate, identifying key determinants of normal organization and castration-induced remodeling. By pinpointing the cell types and molecular programs that confer ADT responsiveness or resistance, our findings offer new directions for prostate cancer modeling and pave the way toward novel therapeutic strategies aimed at enhancing ADT efficacy and preventing the emergence of castration-resistant prostate cancer.

cancer biology↗

bioRxiv2025

Marginal zone B cells are antigenically activated, infiltrate the kidneys, and exacerbate angiotensin II-dependent hypertension in mice

AimsB cells contribute to the development of hypertension, yet, the specific B cell subsets involved, the mechanism underlying their activation, and the relevance of these responses to human disease remain poorly defined. Methods and resultsWe used single-cell RNA sequencing, single-cell B cell receptor (BCR) VDJ sequencing, and high dimensional flow cytometry to characterise B cell responses in murine angiotensin II-induced hypertension. Chronic angiotensin II infusion in male and female mice increased systolic blood pressure and selectively expanded marginal zone B (MZB) cells, with evidence of antigen-dependent activation, including clonal BCR expansion, enrichment of IGHV1 B cell receptor variants, and increased expression of activation markers (CD69 and Nur77). Intercellular communication analyses revealed enhanced antigen-presentation signalling between MZB and CD8+ T cells in hypertensive mice. Activated MZB-like memory B cells also accumulated in the kidneys of hypertensive mice. Consistent with these findings, multiomic analysis of kidneys from patients with hypertensive chronic kidney disease (CKD) demonstrated an increase in memory B cells with a MZB phenotype and enrichment of antigen-presentation-linked communication with CD8+ T cells. Importantly, hypertensive responses to angiotensin II infusion were significantly blunted in mice lacking MZB cells (BAFF-R-/-). ConclusionOur findings identify MZB cells as a selectively activated, antigen-responsive B cell subset that amplified pathogenic immune responses in murine and human hypertension. By linking subset-specific BCR activation to immune cross-talk and disease causality, this study identifies MZB cells - and the (auto)antigens that activate them - as promising targets for precision immunomodulatory strategies in hypertension.

physiology↗

bioRxiv2026

Single-cell characterization and machine learning-based classification reveal the transcriptional identity and antigen-experienced features of peripheral CD4+CD8+ double-positive T cells

T cell mediated immunity depends on diverse subsets with distinct regulatory and cytotoxic roles. While CD4 and CD8 T cells have traditionally been viewed as separate lineages, double-positive T (DPT) cells coexpressing both markers have emerged as a rare subset with potential roles in immune modulation, cytotoxicity, and memory. Here, we integrated single-cell RNA sequencing, single-cell TCR sequencing, and supervised machine learning to characterize transcriptionally defined DPT cells in human peripheral blood. DPT cells formed a reproducible transcriptional state distinct from conventional single-positive T cells and exhibited gene expression programs associated with cytotoxic effector potential, antigen experience, and immune cell migration. Clonal repertoire analysis revealed enrichment of expanded clonotypes and sharing of identical TCR clonotypes with both CD4 and CD8 T cell populations, indicating shared antigen-driven clonal histories rather than a distinct developmental lineage. To enable robust identification of this rare state, we developed a machine learning classifier trained on sorted CD4 and CD8 T cells, which outperformed marker-based approaches. Application of this framework to public COVID-19 single-cell datasets demonstrated its generalizability under immune perturbation. Together, these findings establish DPT cells as a reproducible, antigen-experienced transcriptional state with cytotoxicity- and migration-associated programs, and provide a framework for systematic identification of rare T cell states, expanding our understanding of T cell diversity.

immunology↗

bioRxiv2026

Viral load-driven systemic immune exhaustion is an enabler of antibody breadth in HIV infection

A broadly neutralizing response to a human immunodeficiency virus (HIV) vaccine is a major goal of the field, yet the determinants of antibody breadth remain poorly defined. Antibody responses to HIV evolve over years of infection, developing unusual features such as high mutation rates in the subset of individuals that acquire breadth. Using single-cell RNA sequencing, single-cell proteomics, and plasma neutralization assays in a longitudinal Kenyan cohort of treatment-naive women living with HIV, we identify systems-level immune correlates of antibody breadth. Broad and narrow neutralizers diverge early in infection along a viral load-driven axis with broad neutralizers exhibiting greater viral loads and CD4 T cell decline; concurrently, NK cells, CD8 T cells, and monocytes, broad neutralizers displayed greater magnitude of immune activation in early infection, followed by exhaustion in late infection. This functional decline may dampen NK cell-mediated pruning of T follicular helper cells, creating a permissive environment for sustained germinal center activity and antibody maturation. Narrow neutralizers, by contrast, maintain functional cellular immunity but lack the antigenic pressure and permissive exhaustive state associated with breadth. Together, these findings suggest that antibody breadth in HIV infection reflects failed viral control rather than a successful antiviral response with implications for vaccination and cure strategies that must balance cellular and humoral immunity.

immunology↗

bioRxiv2016

Isoform-level gene expression patterns in single-cell RNA-sequencing data

RNA-sequencing of single-cells enables characterization of transcriptional heterogeneity in seemingly homogenous cell populations. In this study we propose and apply a novel method, ISOform-Patterns (ISOP), based on mixture modeling, to characterize the expression patterns of pairs of isoforms from the same gene in single-cell isoform-level expression data. We define six principal patterns of isoform expression relationships and introduce the concept of differential pattern analysis. We applied ISOP for analysis of single-cell RNA-sequencing data from a breast cancer cell line, with replication in two independent datasets. In the primary dataset we detected and assigned pattern type of 16562 isoform-pairs from 4929 genes. Our results showed that 78% of the isoform pairs displayed a mutually exclusive expression pattern, 14% of the isoform pairs displayed bimodal isoform preference and 8% isoform pairs displayed isoform preference. 26% of the isoform-pair patterns were significant, while remaining isoform-pair patterns can be understood as effects of transcriptional bursting, drop-out and biological heterogeneity. 32% of genes discovered through differential pattern analysis were novel and not detected by differential expression analysis. ISOP provides a novel approach for characterization of isoform-level expression in single-cell populations. Our results reveal a common occurrence of isoform-level preference, commitment and heterogeneity in single-cell populations.

Bioinformatics↗

bioRxiv2021

Single Cell, Single Nucleus and Spatial RNA Sequencing of the Human Liver Identifies Hepatic Stellate Cell and Cholangiocyte Heterogeneity

The critical functions of the human liver are coordinated through the interactions of hepatic parenchymal and non-parenchymal cells. Recent advances in single cell transcriptional approaches have enabled an examination of the human liver with unprecedented resolution. However, dissociation related cell perturbation can limit the ability to fully capture the human livers parenchymal cell fraction, which limits the ability to comprehensively profile this organ. Here, we report the transcriptional landscape of 73,295 cells from the human liver using matched single-cell RNA sequencing (scRNA-seq) and single-nucleus RNA sequencing (snRNA-seq). The addition of snRNA-seq enabled the characterization of interzonal hepatocytes at single-cell resolution, revealed the presence of rare subtypes of hepatic stellate cells previously only seen in disease, and detection of cholangiocyte progenitors that had only been observed during in vitro differentiation experiments. However, T and B lymphocytes and NK cells were only distinguishable using scRNA-seq, highlighting the importance of applying both technologies to obtain a complete map of tissue-resident cell-types. We validated the distinct spatial distribution of the hepatocyte, cholangiocyte and stellate cell populations by an independent spatial transcriptomics dataset and immunohistochemistry. Our study provides a systematic comparison of the transcriptomes captured by scRNA-seq and snRNA-seq and delivers a high-resolution map of the parenchymal cell populations in the healthy human liver.

cell biology↗

bioRxiv2022

SPECK: An Unsupervised Learning Approach for Cell Surface Receptor Abundance Estimation for Single Cell RNA-Sequencing Data

The rapid development of single cell transcriptomics has revolutionized the study of complex tissues. Single cell RNA-sequencing (scRNA-seq) can profile tens-of-thousands of dissociated cells from a tissue sample, enabling researchers to identify cell types, phenotypes and interactions that control tissue structure and function. A key requirement of these applications is the accurate estimation of cell surface protein abundance. Although technologies to directly quantify surface proteins are available, this data is uncommon and limited to proteins with available antibodies. While supervised methods that are trained on Cellular Indexing of Transcriptomes and Epitopes by Sequencing (CITE-seq) data can provide the best performance, this training data is also limited by available antibodies and may not exist for the tissue under investigation. In the absence of protein measurements, researchers must estimate receptor abundance from scRNA-seq data. We thereby developed a new unsupervised method for receptor abundance estimation using scRNA-seq data called SPECK (Surface Protein abundance Estimation using CKmeans-based clustered thresholding) and evaluated its performance against other unsupervised approaches on up to 215 human receptors and multiple tissue types. This analysis reveals that techniques based on a thresholded reduced rank reconstruction (RRR) of scRNA-seq data are effective for receptor abundance estimation with SPECK providing the best overall performance.

bioinformatics↗

bioRxiv2021

Automated annotation of rare-cell types from single-cell RNA-sequencing data through synthetic oversampling

The research landscape of single-cell and single-nuclei RNA sequencing is evolving rapidly, and one area that is enabled by this technology, is the detection of rare cells. An automated, unbiased and accurate annotation of rare subpopulations is challenging. Once rare cells are identified in one dataset, it will usually be necessary to generate other datasets to enrich the analysis (e.g., with samples from other tissues). From a machine learning perspective, the challenge arises from the fact that rare cell subpopulations constitute an imbalanced classification problem. We here introduce a Machine Learning (ML)-based oversampling method that uses gene expression counts of already identified rare cells as an input to generate synthetic cells to then identify similar (rare) cells in other publicly available experiments. We utilize single-cell synthetic oversampling (sc-SynO), which is based on the Localized Random Affine Shadowsampling (LoRAS) algorithm. The algorithm corrects for the overall imbalance ratio of the minority and majority class. We demonstrate the effectiveness of the method for two independent use cases, each consisting of two published datasets. The first use case identifies cardiac glial cells in snRNA-Seq data (17 nuclei out of 8,635). This use case was designed to take a larger imbalance ratio ([~]1 to 500) into account and only uses single-nuclei data. The second use case was designed to jointly use snRNA-Seq data and scRNA-Seq on a lower imbalance ratio ([~]1 to 26) for the training step to likewise investigate the potential of the algorithm to consider both single cell capture procedures and the impact of "less" rare-cell types. For validation purposes, all datasets have also been analyzed in a traditional manner using common data analysis approaches, such as the Seurat3 workflow. Our algorithm identifies rare-cell populations with a high accuracy and low false positive detection rate. A striking benefit of our algorithm is that it can be readily implemented in other and existing workflows. The code basis is publicly available at FairdomHub (https://fairdomhub.org/assays/1368) and can easily be transferred to train other customized approaches.

bioinformatics↗