Search

Increasingly, scRNA-Seq studies explore cell populations across different samples and the effect of sample heterogeneity on organisms phenotype. However, relatively few bioinformatic methods have been developed which adequately address the variation between samples for such population-level analyses. We propose a framework for representing the entire single-cell profile of a sample, which we call a GloScope representation. We implement GloScope on scRNA-Seq datasets from study designs ranging from 12 to over 300 samples and demonstrate how GloScope allows researchers to perform essential bioinformatic tasks at the sample-level, in particular visualization and quality control assessment.

Sample multiplexing enables pooled analysis during single-cell RNA sequencing workflows, thereby increasing throughput and reducing batch effects. A challenge for all multiplexing techniques is to link sample-specific barcodes with cell-specific barcodes, then demultiplex sample identity post-sequencing. However, existing demultiplexing tools fail under many real-world conditions where barcode cross-contamination is an issue. We therefore developed deMULTIplex2, an algorithm inspired by a mechanistic model of barcode cross-contamination. deMULTIplex2 employs generalized linear models and expectation-maximization to probabilistically determine the sample identity of each cell. Benchmarking reveals superior performance across various experimental conditions, particularly on large or noisy datasets with unbalanced sample compositions.

Abstract Increasingly, scRNA-Seq studies explore cell populations across different samples and the effect of sample heterogeneity on organism’s phenotype. However, relatively few bioinformatic methods have been developed which adequately address the variation between samples for such population-level analyses. We propose a framework for representing the entire single-cell profile of a sample, which we call a GloScope representation. We implement GloScope on scRNA-Seq datasets from study designs ranging from 12 to over 300 samples and demonstrate how GloScope allows researchers to perform essential bioinformatic tasks at the sample-level, in particular visualization and quality control assessment.

Abstract Sample multiplexing enables pooled analysis during single-cell RNA sequencing workflows, thereby increasing throughput and reducing batch effects. A challenge for all multiplexing techniques is to link sample-specific barcodes with cell-specific barcodes, then demultiplex sample identity post-sequencing. However, existing demultiplexing tools fail under many real-world conditions where barcode cross-contamination is an issue. We therefore developed deMULTIplex2, an algorithm inspired by a mechanistic model of barcode cross-contamination. deMULTIplex2 employs generalized linear models and expectation–maximization to probabilistically determine the sample identity of each cell. Benchmarking reveals superior performance across various experimental conditions, particularly on large or noisy datasets with unbalanced sample compositions.

Abstract Many deep learning-based methods have been proposed to handle complex single-cell data. Deep learning approaches may also prove useful to jointly analyze single-cell RNA sequencing (scRNA-seq) and single-cell T cell receptor sequencing (scTCR-seq) data for novel discoveries. We developed scNAT, a deep learning method that integrates paired scRNA-seq and scTCR-seq data to represent data in a unified latent space for downstream analysis. We demonstrate that scNAT is capable of removing batch effects, and identifying cell clusters and a T cell migration trajectory from blood to cerebrospinal fluid in multiple sclerosis.

Abstract Imaging-based spatial transcriptomics techniques provide valuable spatial and gene expression information at single-cell resolution. However, their current capability is restricted to profiling a limited number of genes per sample, resulting in most of the transcriptome remaining unmeasured. To overcome this challenge, we develop ENGEP, an ensemble learning-based tool that predicts unmeasured gene expression in spatial transcriptomics data by using multiple single-cell RNA sequencing datasets as references. ENGEP outperforms current state-of-the-art tools and brings biological insight by accurately predicting unmeasured genes. ENGEP has exceptional efficiency in terms of runtime and memory usage, making it scalable for analyzing large datasets.

Abstract Background Somatic embryogenesis is a major process for plant regeneration. However, cell communication and the gene regulatory network responsible for cell reprogramming during somatic embryogenesis are still largely unclear. Recent advances in single-cell technologies enable us to explore the mechanism of plant regeneration at single-cell resolution. Results We generate a high-resolution single-cell transcriptomic landscape of hypocotyl tissue from the highly regenerable cotton genotype Jin668 and the recalcitrant TM-1. We identify nine putative cell clusters and 23 cluster-specific marker genes for both cultivars. We find that the primary vascular cell is the major cell type that undergoes cell fate transition in response to external stimulation. Further developmental trajectory and gene regulatory network analysis of these cell clusters reveals that a total of 41 hormone response-related genes, including LAX2, LAX1, and LOX3, exhibit different expression patterns in the primary xylem and cambium region of Jin668 and TM-1. We also identify novel genes, including CSEF, PIS1, AFB2, ATHB2, PLC2, and PLT3, that are involved in regeneration. We demonstrate that LAX2, LAX1 and LOX3 play important roles in callus proliferation and plant regeneration by CRISPR/Cas9 editing and overexpression assay. Conclusions This study provides novel insights on the role of the regulatory network in cell fate transition and reprogramming during plant regeneration driven by somatic embryogenesis.

Abstract Background Protein annotation is a major goal in molecular biology, yet experimentally determined knowledge is typically limited to a few model organisms. In non-model species, the sequence-based prediction of gene orthology can be used to infer protein identity; however, this approach loses predictive power at longer evolutionary distances. Here we propose a workflow for protein annotation using structural similarity, exploiting the fact that similar protein structures often reflect homology and are more conserved than protein sequences. Results We propose a workflow of openly available tools for the functional annotation of proteins via structural similarity (MorF: MorphologFinder) and use it to annotate the complete proteome of a sponge. Sponges are highly relevant for inferring the early history of animals, yet their proteomes remain sparsely annotated. MorF accurately predicts the functions of proteins with known homology in $${>}90\%$$ > 90 % cases and annotates an additional $$50\%$$ 50 % of the proteome beyond standard sequence-based methods. We uncover new functions for sponge cell types, including extensive FGF, TGF, and Ephrin signaling in sponge epithelia, and redox metabolism and control in myopeptidocytes. Notably, we also annotate genes specific to the enigmatic sponge mesocytes, proposing they function to digest cell walls. Conclusions Our work demonstrates that structural similarity is a powerful approach that complements and extends sequence similarity searches to identify homologous proteins over long evolutionary distances. We anticipate this will be a powerful approach that boosts discovery in numerous -omics datasets, especially for non-model organisms.

Abstract Background Expression quantitative trait loci (eQTL) studies show how genetic variants affect downstream gene expression. Single-cell data allows reconstruction of personalized co-expression networks and therefore the identification of SNPs altering co-expression patterns (co-expression QTLs, co-eQTLs) and the affected upstream regulatory processes using a limited number of individuals. Results We conduct a co-eQTL meta-analysis across four scRNA-seq peripheral blood mononuclear cell datasets using a novel filtering strategy followed by a permutation-based multiple testing approach. Before the analysis, we evaluate the co-expression patterns required for co-eQTL identification using different external resources. We identify a robust set of cell-type-specific co-eQTLs for 72 independent SNPs affecting 946 gene pairs. These co-eQTLs are replicated in a large bulk cohort and provide novel insights into how disease-associated variants alter regulatory networks. One co-eQTL SNP, rs1131017, that is associated with several autoimmune diseases, affects the co-expression of RPS26 with other ribosomal genes. Interestingly, specifically in T cells, the SNP additionally affects co-expression of RPS26 and a group of genes associated with T cell activation and autoimmune disease. Among these genes, we identify enrichment for targets of five T-cell-activation-related transcription factors whose binding sites harbor rs1131017. This reveals a previously overlooked process and pinpoints potential regulators that could explain the association of rs1131017 with autoimmune diseases. Conclusion Our co-eQTL results highlight the importance of studying context-specific gene regulation to understand the biological implications of genetic variation. With the expected growth of sc-eQTL datasets, our strategy and technical guidelines will facilitate future co-eQTL identification, further elucidating unknown disease mechanisms.

The pairing of CRISPR/Cas9-based gene editing with massively parallel single-cell readouts now enables large-scale lineage tracing. However, the rapid growth in complexity of data from these assays has outpaced our ability to accurately infer phylogenetic relationships. First, we introduce Cassiopeia-a suite of scalable maximum parsimony approaches for tree reconstruction. Second, we provide a simulation framework for evaluating algorithms and exploring lineage tracer design principles. Finally, we generate the most complex experimental lineage tracing dataset to date, 34,557 human cells continuously traced over 15 generations, and use it for benchmarking phylogenetic inference approaches. We show that Cassiopeia outperforms traditional methods by several metrics and under a wide variety of parameter regimes, and provide insight into the principles for the design of improved Cas9-enabled recorders. Together, these should broadly enable large-scale mammalian lineage tracing efforts. Cassiopeia and its benchmarking resources are publicly available at www.github.com/YosefLab/Cassiopeia.

Abstract Background The epidermis of cotton ovule produces fibers, the most important natural cellulose source for the global textile industry. However, the molecular mechanism of fiber cell growth is still poorly understood. Results Here, we develop an optimized protoplasting method, and integrate single-cell RNA sequencing (scRNA-seq) and single-cell ATAC sequencing (scATAC-seq) to systematically characterize the cells of the outer integument of ovules from wild type and fuzzless/lintless (fl) cotton (Gossypium hirsutum). By jointly analyzing the scRNA-seq data from wildtype and fl, we identify five cell populations including the fiber cell type and construct the development trajectory for fiber lineage cells. Interestingly, by time-course diurnal transcriptomic analysis, we demonstrate that the primary growth of fiber cells is a highly regulated circadian rhythmic process. Moreover, we identify a small peptide GhRALF1 that circadian rhythmically controls fiber growth possibly through oscillating auxin signaling and proton pump activity in the plasma membrane. Combining with scATAC-seq, we further identify two cardinal cis-regulatory elements (CREs, TCP motif, and TCP-like motif) which are bound by the trans factors GhTCP14s to modulate the circadian rhythmic metabolism of mitochondria and protein translation through regulating approximately one third of genes that are highly expressed in fiber cells. Conclusions We uncover a fiber-specific circadian clock-controlled gene expression program in regulating fiber growth. This study unprecedentedly reveals a new route to improve fiber traits by engineering the circadian clock of fiber cells.

Abstract Background Quality control (QC) of cells, a critical first step in single-cell RNA sequencing data analysis, has largely relied on arbitrarily fixed data-agnostic thresholds applied to QC metrics such as gene complexity and fraction of reads mapping to mitochondrial genes. The few existing data-driven approaches perform QC at the level of samples or studies without accounting for biological variation. Results We first demonstrate that QC metrics vary with both tissue and cell types across technologies, study conditions, and species. We then propose data-driven QC (ddqc), an unsupervised adaptive QC framework to perform flexible and data-driven QC at the level of cell types while retaining critical biological insights and improved power for downstream analysis. ddqc applies an adaptive threshold based on the median absolute deviation on four QC metrics (gene and UMI complexity, fraction of reads mapping to mitochondrial and ribosomal genes). ddqc retains over a third more cells when compared to conventional data-agnostic QC filters. Finally, we show that ddqc recovers biologically meaningful trends in gradation of gene complexity among cell types that can help answer questions of biological interest such as which cell types express the least and most number of transcripts overall, and ribosomal transcripts specifically. Conclusions ddqc retains cell types such as metabolically active parenchymal cells and specialized cells such as neutrophils which are often lost by conventional QC. Taken together, our work proposes a revised paradigm to quality filtering best practices—iterative QC, providing a data-driven QC framework compatible with observed biological diversity.

Abstract Background Cell-cell interactions are important for information exchange between different cells, which are the fundamental basis of many biological processes. Recent advances in single-cell RNA sequencing (scRNA-seq) enable the characterization of cell-cell interactions using computational methods. However, it is hard to evaluate these methods since no ground truth is provided. Spatial transcriptomics (ST) data profiles the relative position of different cells. We propose that the spatial distance suggests the interaction tendency of different cell types, thus could be used for evaluating cell-cell interaction tools. Results We benchmark 16 cell-cell interaction methods by integrating scRNA-seq with ST data. We characterize cell-cell interactions into short-range and long-range interactions using spatial distance distributions between ligands and receptors. Based on this classification, we define the distance enrichment score and apply an evaluation workflow to 16 cell-cell interaction tools using 15 simulated and 5 real scRNA-seq and ST datasets. We also compare the consistency of the results from single tools with the commonly identified interactions. Our results suggest that the interactions predicted by different tools are highly dynamic, and the statistical-based methods show overall better performance than network-based methods and ST-based methods. Conclusions Our study presents a comprehensive evaluation of cell-cell interaction tools for scRNA-seq. CellChat, CellPhoneDB, NicheNet, and ICELLNET show overall better performance than other tools in terms of consistency with spatial tendency and software scalability. We recommend using results from at least two methods to ensure the accuracy of identified interactions. We have packaged the benchmark workflow with detailed documentation at GitHub ( https://github.com/wanglabtongji/CCI ).

Abstract Background The rapid accumulation of single-cell RNA sequencing (scRNA-seq) data presents unique opportunities to decode the genetically mediated cell-type specificity in complex diseases. Here, we develop a new method, scGWAS, which effectively leverages scRNA-seq data to achieve two goals: (1) to infer the cell types in which the disease-associated genes manifest and (2) to construct cellular modules which imply disease-specific activation of different processes. Results scGWAS only utilizes the average gene expression for each cell type followed by virtual search processes to construct the null distributions of module scores, making it scalable to large scRNA-seq datasets. We demonstrated scGWAS in 40 genome-wide association studies (GWAS) datasets (average sample size N ≈ 154,000) using 18 scRNA-seq datasets from nine major human/mouse tissues (totaling 1.08 million cells) and identified 2533 trait and cell-type associations, each with significant modules for further investigation. The module genes were validated using disease or clinically annotated references from ClinVar, OMIM, and pLI variants. Conclusions We showed that the trait-cell type associations identified by scGWAS, while generally constrained to trait-tissue associations, could recapitulate many well-studied relationships and also reveal novel relationships, providing insights into the unsolved trait-tissue associations. Moreover, in each specific cell type, the associations with different traits were often mediated by different sets of risk genes, implying disease-specific activation of driving processes. In summary, scGWAS is a powerful tool for exploring the genetic basis of complex diseases at the cell type level using single-cell expression data.

Abstract RNA processing, including splicing and alternative polyadenylation, is crucial to gene function and regulation, but methods to detect RNA processing from single-cell RNA sequencing data are limited by reliance on pre-existing annotations, peak calling heuristics, and collapsing measurements by cell type. We introduce ReadZS, an annotation-free statistical approach to identify regulated RNA processing in single cells. ReadZS discovers cell type-specific RNA processing in human lung and conserved, developmentally regulated RNA processing in mammalian spermatogenesis—including global 3′ UTR shortening in human spermatogenesis. ReadZS also discovers global 3′ UTR lengthening in Arabidopsis development, highlighting the usefulness of this method in under-annotated transcriptomes.

scRNA-seq dataset integration occurs in different contexts, such as the identification of cell type-specific differences in gene expression across conditions or species, or batch effect correction. We present scAlign, an unsupervised deep learning method for data integration that can incorporate partial, overlapping, or a complete set of cell labels, and estimate per-cell differences in gene expression across datasets. scAlign performance is state-of-the-art and robust to cross-dataset variation in cell type-specific expression and cell type composition. We demonstrate that scAlign reveals gene expression programs for rare populations of malaria parasites. Our framework is widely applicable to integration challenges in other domains.

Technical variation in feature measurements, such as gene expression and locus accessibility, is a key challenge of large-scale single-cell genomic datasets. We show that this technical variation in both scRNA-seq and scATAC-seq datasets can be mitigated by analyzing feature detection patterns alone and ignoring feature quantification measurements. This result holds when datasets have low detection noise relative to quantification noise. We demonstrate state-of-the-art performance of detection pattern models using our new framework, scBFA, for both cell type identification and trajectory inference. Performance gains can also be realized in one line of R code in existing pipelines.

Abstract It is a challenging task to integrate scRNA-seq and scATAC-seq data obtained from different batches. Existing methods tend to use a pre-defined gene activity matrix to convert the scATAC-seq data into scRNA-seq data. The pre-defined gene activity matrix is often of low quality and does not reflect the dataset-specific relationship between the two data modalities. We propose scDART, a deep learning framework that integrates scRNA-seq and scATAC-seq data and learns cross-modalities relationships simultaneously. Specifically, the design of scDART allows it to preserve cell trajectories in continuous cell populations and can be applied to trajectory inference on integrated data.

Abstract A challenge in bulk gene differential expression analysis is to differentiate changes due to cell type-specific gene expression and cell type proportions. SCADIE is an iterative algorithm that simultaneously estimates cell type-specific gene expression profiles and cell type proportions, and performs cell type-specific differential expression analysis at the group level. Through its unique penalty and objective function, SCADIE more accurately identifies cell type-specific differentially expressed genes than existing methods, including those that may be missed from single cell RNA-Seq data. SCADIE has robust performance with respect to the choice of deconvolution methods and the sources and quality of input data.

Abstract Background Multiplexing of samples in single-cell RNA-seq studies allows a significant reduction of the experimental costs, straightforward identification of doublets, increased cell throughput, and reduction of sample-specific batch effects. Recently published multiplexing techniques using oligo-conjugated antibodies or -lipids allow barcoding sample-specific cells, a process called “hashing.” Results Here, we compare the hashing performance of TotalSeq-A and -C antibodies, custom synthesized lipids and MULTI-seq lipid hashes in four cell lines, both for single-cell RNA-seq and single-nucleus RNA-seq. We also compare TotalSeq-B antibodies with CellPlex reagents (10x Genomics) on human PBMCs and TotalSeq-B with different lipids on primary mouse tissues. Hashing efficiency was evaluated using the intrinsic genetic variation of the cell lines and mouse strains. Antibody hashing was further evaluated on clinical samples using PBMCs from healthy and SARS-CoV-2 infected patients, where we demonstrate a more affordable approach for large single-cell sequencing clinical studies, while simultaneously reducing batch effects. Conclusions Benchmarking of different hashing strategies and computational pipelines indicates that correct demultiplexing can be achieved with both lipid- and antibody-hashed human cells and nuclei, with MULTISeqDemux as the preferred demultiplexing function and antibody-based hashing as the most efficient protocol on cells. On nuclei datasets, lipid hashing delivers the best results. Lipid hashing also outperforms antibodies on cells isolated from mouse brain. However, antibodies demonstrate better results on tissues like spleen or lung.

Abstract The ability to discover new cell phenotypes by unsupervised clustering of single-cell transcriptomes has revolutionized biology. Currently, there is no principled way to decide whether a cluster of cells contains meaningful subpopulations that should be further resolved. Here, we present phiclust (ϕ clust ), a clusterability measure derived from random matrix theory that can be used to identify cell clusters with non-random substructure, testably leading to the discovery of previously overlooked phenotypes.

Abstract We consider an increasingly popular study design where single-cell RNA-seq data are collected from multiple individuals and the question of interest is to find genes that are differentially expressed between two groups of individuals. Towards this end, we propose a statistical method named IDEAS (individual level differential expression analysis for scRNA-seq). For each gene, IDEAS summarizes its expression in each individual by a distribution and then assesses whether these individual-specific distributions are different between two groups of individuals. We apply IDEAS to assess gene expression differences of autism patients versus controls and COVID-19 patients with mild versus severe symptoms.

Abstract Background Single-cell RNA-sequencing (scRNA-seq) technologies and associated analysis methods have rapidly developed in recent years. This includes preprocessing methods, which assign sequencing reads to genes to create count matrices for downstream analysis. While several packaged preprocessing workflows have been developed to provide users with convenient tools for handling this process, how they compare to one another and how they influence downstream analysis have not been well studied. Results Here, we systematically benchmark the performance of 10 end-to-end preprocessing workflows (Cell Ranger, Optimus, salmon alevin, alevin-fry, kallisto bustools, dropSeqPipe, scPipe, zUMIs, celseq2, and scruff) using datasets yielding different biological complexity levels generated by CEL-Seq2 and 10x Chromium platforms. We compare these workflows in terms of their quantification properties directly and their impact on normalization and clustering by evaluating the performance of different method combinations. While the scRNA-seq preprocessing workflows compared vary in their detection and quantification of genes across datasets, after downstream analysis with performant normalization and clustering methods, almost all combinations produce clustering results that agree well with the known cell type labels that provided the ground truth in our analysis. Conclusions In summary, the choice of preprocessing method was found to be less important than other steps in the scRNA-seq analysis process. Our study comprehensively compares common scRNA-seq preprocessing workflows and summarizes their characteristics to guide workflow users.

Abstract Background Cell type identification is one of the most important questions in single-cell RNA sequencing (scRNA-seq) data analysis. With the accumulation of public scRNA-seq data, supervised cell type identification methods have gained increasing popularity due to better accuracy, robustness, and computational performance. Despite all the advantages, the performance of the supervised methods relies heavily on several key factors: feature selection, prediction method, and, most importantly, choice of the reference dataset. Results In this work, we perform extensive real data analyses to systematically evaluate these strategies in supervised cell identification. We first benchmark nine classifiers along with six feature selection strategies and investigate the impact of reference data size and number of cell types in cell type prediction. Next, we focus on how discrepancies between reference and target datasets and how data preprocessing such as imputation and batch effect correction affect prediction performance. We also investigate the strategies of pooling and purifying reference data. Conclusions Based on our analysis results, we provide guidelines for using supervised cell typing methods. We suggest combining all individuals from available datasets to construct the reference dataset and use multi-layer perceptron (MLP) as the classifier, along with F-test as the feature selection method. All the code used for our analysis is available on GitHub ( https://github.com/marvinquiet/RefConstruction_supervisedCelltyping ).

Abstract Single-cell RNA-seq (scRNA-seq) profiles gene expression with high resolution. Here, we develop a stepwise computational method-called SCAPTURE to identify, evaluate, and quantify cleavage and polyadenylation sites (PASs) from 3′ tag-based scRNA-seq. SCAPTURE detects PASs de novo in single cells with high sensitivity and accuracy, enabling detection of previously unannotated PASs. Quantified alternative PAS transcripts refine cell identity analysis beyond gene expression, enriching information extracted from scRNA-seq data. Using SCAPTURE, we show changes of PAS usage in PBMCs from infected versus healthy individuals at single-cell resolution.

Abstract Background Many neurodegenerative diseases develop only later in life, when cells in the nervous system lose their structure or function. In many forms of neurodegenerative diseases, this late-onset phenomenon remains largely unexplained. Results Analyzing single-cell RNA sequencing from Alzheimer’s disease (AD) and Huntington’s disease (HD) patients, we find increased transcriptional heterogeneity in disease-state neurons. We hypothesize that transcriptional heterogeneity precedes neurodegenerative disease pathologies. To test this idea experimentally, we use juvenile forms (72Q; 180Q) of HD iPSCs, differentiate them into committed neuronal progenitors, and obtain single-cell expression profiles. We show a global increase in gene expression variability in HD. Autophagy genes become more stable, while energy and actin-related genes become more variable in the mutant cells. Knocking down several differentially variable genes results in increased aggregate formation, a pathology associated with HD. We further validate the increased transcriptional heterogeneity in CHD8+/− cells, a model for autism spectrum disorder. Conclusions Overall, our results suggest that although neurodegenerative diseases develop over time, transcriptional regulation imbalance is present already at very early developmental stages. Therefore, an intervention aimed at this early phenotype may be of high diagnostic value.

Abstract The integration of single-cell RNA-sequencing datasets from multiple sources is critical for deciphering cell-to-cell heterogeneities and interactions in complex biological systems. We present a novel unsupervised batch effect removal framework, called iMAP, based on both deep autoencoders and generative adversarial networks. Compared with current methods, iMAP shows superior, robust, and scalable performance in terms of both reliably detecting the batch-specific cells and effectively mixing distributions of the batch-shared cell types. Applying iMAP to tumor microenvironment datasets from two platforms, Smart-seq2 and 10x Genomics, we find that iMAP can leverage the powers of both platforms to discover novel cell-cell interactions.

Abstract We present a Minimal Event Distance Aneuploidy Lineage Tree (MEDALT) algorithm that infers the evolution history of a cell population based on single-cell copy number (SCCN) profiles, and a statistical routine named lineage speciation analysis (LSA), whichty facilitates discovery of fitness-associated alterations and genes from SCCN lineage trees. MEDALT appears more accurate than phylogenetics approaches in reconstructing copy number lineage. From data from 20 triple-negative breast cancer patients, our approaches effectively prioritize genes that are essential for breast cancer cell fitness and predict patient survival, including those implicating convergent evolution. The source code of our study is available at https://github.com/KChen-lab/MEDALT .

Abstract High-throughput single-cell RNA-seq (scRNA-seq) is a powerful tool for studying gene expression in single cells. Most current scRNA-seq bioinformatics tools focus on analysing overall expression levels, largely ignoring alternative mRNA isoform expression. We present a computational pipeline, Sierra, that readily detects differential transcript usage from data generated by commonly used polyA-captured scRNA-seq technology. We validate Sierra by comparing cardiac scRNA-seq cell types to bulk RNA-seq of matched populations, finding significant overlap in differential transcripts. Sierra detects differential transcript usage across human peripheral blood mononuclear cells and the Tabula Muris, and 3 ′UTR shortening in cardiac fibroblasts. Sierra is available at https://github.com/VCCRI/Sierra .

Abstract Background Single-cell RNA sequencing has been widely adopted to estimate the cellular composition of heterogeneous tissues and obtain transcriptional profiles of individual cells. Multiple approaches for optimal sample dissociation and storage of single cells have been proposed as have single-nuclei profiling methods. What has been lacking is a systematic comparison of their relative biases and benefits. Results Here, we compare gene expression and cellular composition of single-cell suspensions prepared from adult mouse kidney using two tissue dissociation protocols. For each sample, we also compare fresh cells to cryopreserved and methanol-fixed cells. Lastly, we compare this single-cell data to that generated using three single-nucleus RNA sequencing workflows. Our data confirms prior reports that digestion on ice avoids the stress response observed with 37 °C dissociation. It also reveals cell types more abundant either in the cold or warm dissociations that may represent populations that require gentler or harsher conditions to be released intact. For cell storage, cryopreservation of dissociated cells results in a major loss of epithelial cell types; in contrast, methanol fixation maintains the cellular composition but suffers from ambient RNA leakage. Finally, cell type composition differences are observed between single-cell and single-nucleus RNA sequencing libraries. In particular, we note an underrepresentation of T, B, and NK lymphocytes in the single-nucleus libraries. Conclusions Systematic comparison of recovered cell types and their transcriptional profiles across the workflows has highlighted protocol-specific biases and thus enables researchers starting single-cell experiments to make an informed choice.

Abstract The pairing of CRISPR/Cas9-based gene editing with massively parallel single-cell readouts now enables large-scale lineage tracing. However, the rapid growth in complexity of data from these assays has outpaced our ability to accurately infer phylogenetic relationships. First, we introduce Cassiopeia—a suite of scalable maximum parsimony approaches for tree reconstruction. Second, we provide a simulation framework for evaluating algorithms and exploring lineage tracer design principles. Finally, we generate the most complex experimental lineage tracing dataset to date, 34,557 human cells continuously traced over 15 generations, and use it for benchmarking phylogenetic inference approaches. We show that Cassiopeia outperforms traditional methods by several metrics and under a wide variety of parameter regimes, and provide insight into the principles for the design of improved Cas9-enabled recorders. Together, these should broadly enable large-scale mammalian lineage tracing efforts. Cassiopeia and its benchmarking resources are publicly available at www.github.com/YosefLab/Cassiopeia .

Abstract Background Early single-cell RNA-seq (scRNA-seq) studies suggested that it was unusual to see more than one isoform being produced from a gene in a single cell, even when multiple isoforms were detected in matched bulk RNA-seq samples. However, these studies generally did not consider the impact of dropouts or isoform quantification errors, potentially confounding the results of these analyses. Results In this study, we take a simulation based approach in which we explicitly account for dropouts and isoform quantification errors. We use our simulations to ask to what extent it is possible to study alternative splicing using scRNA-seq. Additionally, we ask what limitations must be overcome to make splicing analysis feasible. We find that the high rate of dropouts associated with scRNA-seq is a major obstacle to studying alternative splicing. In mice and other well-established model organisms, the relatively low rate of isoform quantification errors poses a lesser obstacle to splicing analysis. We find that different models of isoform choice meaningfully change our simulation results. Conclusions To accurately study alternative splicing with single-cell RNA-seq, a better understanding of isoform choice and the errors associated with scRNA-seq is required. An increase in the capture efficiency of scRNA-seq would also be beneficial. Until some or all of the above are achieved, we do not recommend attempting to resolve isoforms in individual cells using scRNA-seq.

Abstract Droplet-based microfluidic devices have become widely used to perform single-cell RNA sequencing (scRNA-seq). However, ambient RNA present in the cell suspension can be aberrantly counted along with a cell’s native mRNA and result in cross-contamination of transcripts between different cell populations. DecontX is a novel Bayesian method to estimate and remove contamination in individual cells. DecontX accurately predicts contamination levels in a mouse-human mixture dataset and removes aberrant expression of marker genes in PBMC datasets. We also compare the contamination levels between four different scRNA-seq protocols. Overall, DecontX can be incorporated into scRNA-seq workflows to improve downstream analyses.

Abstract Background Many functional analysis tools have been developed to extract functional and mechanistic insight from bulk transcriptome data. With the advent of single-cell RNA sequencing (scRNA-seq), it is in principle possible to do such an analysis for single cells. However, scRNA-seq data has characteristics such as drop-out events and low library sizes. It is thus not clear if functional TF and pathway analysis tools established for bulk sequencing can be applied to scRNA-seq in a meaningful way. Results To address this question, we perform benchmark studies on simulated and real scRNA-seq data. We include the bulk-RNA tools PROGENy, GO enrichment, and DoRothEA that estimate pathway and transcription factor (TF) activities, respectively, and compare them against the tools SCENIC/AUCell and metaVIPER, designed for scRNA-seq. For the in silico study, we simulate single cells from TF/pathway perturbation bulk RNA-seq experiments. We complement the simulated data with real scRNA-seq data upon CRISPR-mediated knock-out. Our benchmarks on simulated and real data reveal comparable performance to the original bulk data. Additionally, we show that the TF and pathway activities preserve cell type-specific variability by analyzing a mixture sample sequenced with 13 scRNA-seq protocols. We also provide the benchmark data for further use by the community. Conclusions Our analyses suggest that bulk-based functional analysis tools that use manually curated footprint gene sets can be applied to scRNA-seq data, partially outperforming dedicated single-cell tools. Furthermore, we find that the performance of functional analysis tools is more sensitive to the gene sets than to the statistic used.

Abstract A variety of methods have been developed to demultiplex pooled samples in a single cell RNA sequencing (scRNA-seq) experiment which either require hashtag barcodes or sample genotypes prior to pooling. We introduce scSplit which utilizes genetic differences inferred from scRNA-seq data alone to demultiplex pooled samples. scSplit also enables mapping clusters to original samples. Using simulated, merged, and pooled multi-individual datasets, we show that scSplit prediction is highly concordant with demuxlet predictions and is highly consistent with the known truth in cell-hashing dataset. scSplit is ideally suited to samples without external genotype information and is available at: https://github.com/jon-xu/scSplit

Abstract scRNA-seq profiles each represent a highly partial sample of mRNA molecules from a unique cell that can never be resampled, and robust analysis must separate the sampling effect from biological variance. We describe a methodology for partitioning scRNA-seq datasets into metacells: disjoint and homogenous groups of profiles that could have been resampled from the same cell. Unlike clustering analysis, our algorithm specializes at obtaining granular as opposed to maximal groups. We show how to use metacells as building blocks for complex quantitative transcriptional maps while avoiding data smoothing. Our algorithms are implemented in the MetaCell R/C++ software package.

Abstract Background Single-cell transcriptomics is rapidly advancing our understanding of the cellular composition of complex tissues and organisms. A major limitation in most analysis pipelines is the reliance on manual annotations to determine cell identities, which are time-consuming and irreproducible. The exponential growth in the number of cells and samples has prompted the adaptation and development of supervised classification methods for automatic cell identification. Results Here, we benchmarked 22 classification methods that automatically assign cell identities including single-cell-specific and general-purpose classifiers. The performance of the methods is evaluated using 27 publicly available single-cell RNA sequencing datasets of different sizes, technologies, species, and levels of complexity. We use 2 experimental setups to evaluate the performance of each method for within dataset predictions (intra-dataset) and across datasets (inter-dataset) based on accuracy, percentage of unclassified cells, and computation time. We further evaluate the methods’ sensitivity to the input features, number of cells per population, and their performance across different annotation levels and datasets. We find that most classifiers perform well on a variety of datasets with decreased accuracy for complex datasets with overlapping classes or deep annotations. The general-purpose support vector machine classifier has overall the best performance across the different experiments. Conclusions We present a comprehensive evaluation of automatic cell identification methods for single-cell RNA sequencing data. All the code used for the evaluation is available on GitHub (https://github.com/tabdelaal/scRNAseq_Benchmark). Additionally, we provide a Snakemake workflow to facilitate the benchmarking and to support the extension of new methods and new datasets.

Abstract Technical variation in feature measurements, such as gene expression and locus accessibility, is a key challenge of large-scale single-cell genomic datasets. We show that this technical variation in both scRNA-seq and scATAC-seq datasets can be mitigated by analyzing feature detection patterns alone and ignoring feature quantification measurements. This result holds when datasets have low detection noise relative to quantification noise. We demonstrate state-of-the-art performance of detection pattern models using our new framework, scBFA, for both cell type identification and trajectory inference. Performance gains can also be realized in one line of R code in existing pipelines.

Abstract scRNA-seq dataset integration occurs in different contexts, such as the identification of cell type-specific differences in gene expression across conditions or species, or batch effect correction. We present scAlign, an unsupervised deep learning method for data integration that can incorporate partial, overlapping, or a complete set of cell labels, and estimate per-cell differences in gene expression across datasets. scAlign performance is state-of-the-art and robust to cross-dataset variation in cell type-specific expression and cell type composition. We demonstrate that scAlign reveals gene expression programs for rare populations of malaria parasites. Our framework is widely applicable to integration challenges in other domains.

Abstract Background In high-throughput studies, hundreds to millions of hypotheses are typically tested. Statistical methods that control the false discovery rate (FDR) have emerged as popular and powerful tools for error rate control. While classic FDR methods use only p values as input, more modern FDR methods have been shown to increase power by incorporating complementary information as informative covariates to prioritize, weight, and group hypotheses. However, there is currently no consensus on how the modern methods compare to one another. We investigate the accuracy, applicability, and ease of use of two classic and six modern FDR-controlling methods by performing a systematic benchmark comparison using simulation studies as well as six case studies in computational biology. Results Methods that incorporate informative covariates are modestly more powerful than classic approaches, and do not underperform classic approaches, even when the covariate is completely uninformative. The majority of methods are successful at controlling the FDR, with the exception of two modern methods under certain settings. Furthermore, we find that the improvement of the modern FDR methods over the classic methods increases with the informativeness of the covariate, total number of hypothesis tests, and proportion of truly non-null hypotheses. Conclusions Modern FDR methods that use an informative covariate provide advantages over classic FDR-controlling procedures, with the relative gain dependent on the application and informativeness of available covariates. We present our findings as a practical guide and provide recommendations to aid researchers in their choice of methods to correct for false discoveries.

Abstract Recent technical improvements in single-cell RNA sequencing (scRNA-seq) have enabled massively parallel profiling of transcriptomes, thereby promoting large-scale studies encompassing a wide range of cell types of multicellular organisms. With this background, we propose CellFishing.jl, a new method for searching atlas-scale datasets for similar cells and detecting noteworthy genes of query cells with high accuracy and throughput. Using multiple scRNA-seq datasets, we validate that our method demonstrates comparable accuracy to and is markedly faster than the state-of-the-art software. Moreover, CellFishing.jl is scalable to more than one million cells, and the throughput of the search is approximately 1600 cells per second.

Abstract Single-cell RNA-seq has the potential to facilitate isoform quantification as the confounding factor of a mixed population of cells is eliminated. However, best practice for using existing quantification methods has not been established. We carry out a benchmark for five popular isoform quantification tools. Performance is generally good for simulated data based on SMARTer and SMART-seq2 data. The reduction in performance compared with bulk RNA-seq is small. An important biological insight comes from our analysis of real data which shows that genes that express two isoforms in bulk RNA-seq predominantly express one or neither isoform in individual cells.

Abstract Single-cell analysis is a powerful tool for dissecting the cellular composition within a tissue or organ. However, it remains difficult to detect rare and common cell types at the same time. Here, we present a new computational method, GiniClust2, to overcome this challenge. GiniClust2 combines the strengths of two complementary approaches, using the Gini index and Fano factor, respectively, through a cluster-aware, weighted ensemble clustering technique. GiniClust2 successfully identifies both common and rare cell types in diverse datasets, outperforming existing methods. GiniClust2 is scalable to large datasets.