Single-Cell Sequencing in Bioinformatics

  • Single-cell sequencing is an advanced genomic technology that enables researchers to investigate genetic material, gene expression, chromatin accessibility, and other molecular characteristics at the level of individual cells. Unlike conventional bulk sequencing, which analyses DNA or RNA collected from a large population of cells, single-cell sequencing examines individual cells separately or assigns molecular measurements to individual cells using unique barcodes. This approach makes it possible to identify cellular differences that may otherwise be hidden when molecular information from thousands or millions of cells is combined. Single-cell sequencing has become an important field in bioinformatics, molecular biology, cancer research, immunology, developmental biology, neuroscience, stem-cell research, and precision medicine.
  • Biological tissues contain different cell types, developmental stages, and functional states. Even cells belonging to the same tissue or apparently identical cell population can differ in gene expression, genetic mutations, regulatory activity, and responses to environmental conditions. Bulk sequencing produces an average molecular profile across the cells in a sample, potentially concealing rare cell populations and important differences between individual cells. Single-cell sequencing addresses this limitation by preserving information about individual cells, allowing researchers to investigate cellular heterogeneity, identify rare populations, and understand how cells change during development, disease, and treatment.
  • The term single-cell sequencing describes several related technologies rather than one specific experimental method. Single-cell RNA sequencing (scRNA-seq) measures gene expression in individual cells by analysing RNA molecules, while single-cell DNA sequencing investigates genomic variation within individual cells. Single-cell ATAC sequencing examines chromatin accessibility and regulatory regions. Other approaches investigate DNA methylation, chromatin structure, protein expression, immune-cell receptors, or multiple molecular features simultaneously. The appropriate method depends on the biological question, the type of sample, and the molecular information required.
  • Single-cell RNA sequencing is one of the most widely used single-cell technologies. It measures RNA transcripts in individual cells and provides information about which genes are expressed, how strongly they are expressed, and how expression patterns differ among cell populations. This makes it possible to identify cell types, investigate cellular activation, study developmental trajectories, and characterize changes associated with disease. For example, scRNA-seq can distinguish different immune-cell populations within a tumour, identify rare cell states, and investigate how cells respond to treatment.
  • The first stage of a single-cell sequencing experiment is sample preparation. Researchers collect biological material such as blood, tissue, tumour samples, organoids, cultured cells, or other specimens. The sample is processed to obtain individual cells or, in some applications, isolated cell nuclei. Tissue samples may require mechanical or enzymatic dissociation to release individual cells. The resulting suspension is filtered and assessed to remove debris and large aggregates. Cell viability, concentration, and sample quality are important because damaged cells can release RNA and other cellular material that may interfere with downstream analysis.
  • Cell isolation is a critical step in many single-cell workflows. Individual cells can be separated using fluorescence-activated cell sorting (FACS), microfluidic devices, microwell systems, or droplet-based technologies. FACS uses fluorescent markers and flow cytometry to identify and sort selected cells. Microfluidic systems manipulate small volumes of liquid and can isolate cells into separate reaction compartments. Microwell approaches capture cells in small physical wells, while droplet-based systems encapsulate individual cells in tiny oil droplets containing molecular barcodes and other reagents.
  • Not every single-cell experiment requires intact cells. Single-nucleus sequencing isolates nuclei from biological material rather than separating complete cells. This approach is particularly useful for frozen tissues, difficult-to-dissociate samples, and tissues in which enzymatic processing could alter gene expression or selectively damage particular cell types. Single-nucleus RNA sequencing (snRNA-seq) measures RNA associated with isolated nuclei and can provide information about nuclear transcripts, including pre-mRNA. However, its expression profiles may differ from those obtained from whole-cell RNA sequencing because much of the mature RNA is located in the cytoplasm.
  • In droplet-based single-cell RNA sequencing, individual cells are combined with small particles or beads carrying cell-specific molecular barcodes. Each bead contains many copies of barcode sequences and unique molecular identifiers (UMIs). When a cell and a barcode-containing bead are captured within the same droplet, cellular RNA is released and converted into complementary DNA (cDNA). The barcode identifies the cell of origin, while the UMI helps identify individual original RNA molecules. After reverse transcription and library preparation, sequencing reads can be assigned to their corresponding cells and molecular identifiers.
  • Cell barcodes and UMIs are fundamental components of many single-cell sequencing workflows. A cell barcode identifies the cell or cellular compartment from which a transcript originated. A UMI is a short sequence attached to an original RNA molecule or its derived cDNA, enabling computational methods to distinguish original molecules from PCR duplicates. This helps reduce amplification-related counting bias and improves the estimation of transcript abundance. However, barcode errors, sequencing errors, barcode collisions, and imperfect molecular capture can still affect data quality.
  • An alternative approach uses microwells or plate-based methods to isolate cells individually. In plate-based workflows, cells are deposited into separate wells, and molecular reactions are performed independently. Some methods can capture a larger proportion of the RNA from each cell than certain high-throughput droplet workflows, but they may process fewer cells and require more individual handling. The choice between droplet-based, plate-based, and other isolation methods depends on the desired number of cells, sensitivity, cost, sample type, and experimental objectives.
  • After cell isolation and RNA capture, messenger RNA is usually converted into cDNA through reverse transcription. The cDNA is amplified and prepared for sequencing by adding the necessary adapters and sample indexes. Some protocols capture the full length of transcripts, while others focus mainly on the 3′ or 5′ ends of RNA molecules. Full-length methods can provide more information about transcript structure and alternative splicing, whereas end-counting approaches are often designed for efficient gene-expression quantification across large numbers of cells. The selected library preparation strategy influences the type of biological information that can be recovered.
  • Sequencing is commonly performed using short-read platforms, particularly Illumina systems, although other sequencing technologies can be used in specialized applications. The sequencing output is typically stored in FASTQ files containing nucleotide sequences and quality scores. Reads may include cell barcodes, UMIs, transcript-derived sequences, and sample indexes, depending on the library design. The sequencing depth and number of cells influence the sensitivity of the experiment. Sequencing more reads per cell can improve transcript detection, but increasing the number of cells may be more useful when the objective is to discover rare cell populations.
  • Single-cell RNA sequencing data require specialized bioinformatics processing because each read must be associated with a cell barcode and, where applicable, a UMI. A typical computational workflow begins with raw-read quality control, barcode processing, UMI correction, alignment or transcript quantification, and the generation of a gene-by-cell count matrix. This matrix contains the number of detected RNA molecules assigned to each gene in each cell. The matrix is the starting point for many downstream analyses, including normalization, dimensionality reduction, clustering, cell-type identification, and differential gene-expression analysis.
  • Several software tools are widely used for processing single-cell sequencing data. Cell Ranger is commonly used in workflows associated with Chromium single-cell experiments. STARsolo provides a method for processing barcode-based single-cell RNA sequencing data using the STAR aligner. Alevin-fry and related tools support efficient transcript quantification and barcode processing. Depending on the assay, other tools may be required for barcode correction, read alignment, transcript assignment, and UMI counting. The choice of software should match the library chemistry, barcode structure, reference genome, and sequencing protocol.
  • An illustrative Cell Ranger command for processing a compatible single-cell gene-expression library is:

cellranger count \

  –id=sample_analysis \

  –transcriptome=reference_transcriptome \

  –fastqs=fastq_directory \

  –sample=sample_name

  • This example assumes that a compatible reference transcriptome and FASTQ files have been prepared. The required parameters and workflow depend on the assay and software version. The output commonly includes a filtered gene-expression matrix, barcode information, feature information, and quality-control summaries.
  • Quality control is essential in single-cell sequencing because individual cells may differ considerably in their technical quality. Common quality metrics include the number of detected genes per cell, total UMI counts, percentage of mitochondrial transcripts, and the proportion of reads assigned to valid cell barcodes. Cells with very few detected genes may represent damaged or low-quality cells, while cells with unusually high gene counts may represent doublets or multiplets. A high proportion of mitochondrial transcripts can indicate cellular stress or damage, although appropriate thresholds vary among tissues and cell types.
  • Quality control also includes identifying doublets, which occur when two or more cells are captured or processed together but are mistakenly treated as a single cell. Doublets can produce artificial gene-expression profiles containing markers from different cell types. Tools such as Scrublet, DoubletFinder, and scDblFinder can help identify likely doublets. These tools estimate doublet probability rather than providing definitive experimental proof, so their results should be interpreted alongside sample preparation, cell loading, and biological information.
  • The gene-by-cell count matrix is usually normalized before comparing gene expression among cells. Normalization accounts for differences in sequencing depth and the total number of captured molecules per cell. Log transformation is commonly applied to normalized counts to reduce the influence of highly expressed genes. Highly variable genes are then identified because they contain informative variation across cells. These genes are frequently used for dimensionality reduction and clustering. However, normalization methods and feature selection can influence downstream results, so the analytical strategy should be appropriate for the experimental design and biological question.
  • Dimensionality reduction is used to represent complex gene-expression profiles in a smaller number of dimensions while preserving important patterns of variation. Principal component analysis (PCA) is frequently used as an initial dimensionality-reduction method. Nonlinear methods such as t-distributed stochastic neighbor embedding (t-SNE) and Uniform Manifold Approximation and Projection (UMAP) are often used to visualize cellular relationships in two-dimensional or three-dimensional spaces. These visualizations can reveal groups of cells with similar expression profiles, but distances and cluster separation in a low-dimensional plot should not automatically be interpreted as precise biological distances.
  • Clustering is an important step in single-cell analysis. It groups cells according to similarities in their molecular profiles and can help identify distinct cell populations or states. Graph-based clustering methods are commonly used after dimensionality reduction and construction of a cell-neighbourhood graph. Popular analytical frameworks include Seurat in R and Scanpy in Python. These tools support data normalization, feature selection, dimensionality reduction, clustering, visualization, and other downstream analyses. Clustering results depend on parameters such as the number of principal components, neighbourhood size, and clustering resolution. Therefore, clusters should be evaluated using biological knowledge and independent evidence rather than accepted solely because they appear as separate groups in a visualization.
  • An illustrative Python workflow using Scanpy is:

import scanpy as sc

# Load a single-cell gene-expression matrix

adata = sc.read_10x_mtx(

    “filtered_feature_bc_matrix/”,

    var_names=”gene_symbols”,

    cache=True

)

# Calculate basic quality-control metrics

sc.pp.calculate_qc_metrics(

    adata,

    inplace=True

)

# Filter low-quality cells and genes

sc.pp.filter_cells(adata, min_genes=200)

sc.pp.filter_genes(adata, min_cells=3)

# Normalize and log-transform expression values

sc.pp.normalize_total(adata, target_sum=10000)

sc.pp.log1p(adata)

# Identify highly variable genes

sc.pp.highly_variable_genes(

    adata,

    min_mean=0.0125,

    max_mean=3,

    min_disp=0.5

)

# Retain highly variable genes

adata = adata[:, adata.var.highly_variable].copy()

# Scale data and perform dimensionality reduction

sc.pp.scale(adata, max_value=10)

sc.tl.pca(adata)

# Construct a cell-neighbourhood graph

sc.pp.neighbors(adata, n_neighbors=10, n_pcs=40)

# Perform clustering and visualization

sc.tl.leiden(adata)

sc.tl.umap(adata)

# Display the UMAP representation

sc.pl.umap(adata, color=[“leiden”])

  • This is an illustrative workflow rather than a complete analysis pipeline. In a real experiment, researchers should perform appropriate quality filtering, account for mitochondrial and other assay-specific metrics, assess doublets, and choose suitable normalization and integration methods. Reference genome and annotation versions should also be recorded.
  • After clustering, cell types are identified through marker-gene analysis. Marker genes are genes that show characteristic expression patterns in particular cell types or cellular states. For example, immune-cell populations may be distinguished using combinations of genes associated with T cells, B cells, natural killer cells, monocytes, and macrophages. Researchers compare the expression patterns of candidate markers with published literature, reference atlases, and curated cell-type databases. Automated annotation tools can support this process, but predicted labels should be checked because related cell types may share markers or display different expression patterns under particular biological conditions.
  • Cell-type annotation can be performed using reference-based and marker-based approaches. Reference-based methods compare the expression profiles of unknown cells with previously annotated reference datasets. Tools such as SingleR, CellTypist, and scmap support different forms of automated cell-type classification. Marker-based annotation relies on known biological markers and expert interpretation. Reference-based predictions can be affected by differences in tissue source, disease state, developmental stage, sequencing technology, and reference dataset composition. Therefore, cell-type labels should include appropriate uncertainty where the available evidence is insufficient.
  • Differential gene-expression analysis is used to identify genes whose expression differs between cell populations, conditions, or experimental groups. For example, researchers may compare immune cells from healthy individuals and patients with a disease, or tumour cells collected before and after treatment. Differential expression can reveal genes associated with cellular activation, inflammation, metabolism, proliferation, stress responses, and other biological processes. However, individual cells from the same biological sample are not independent biological replicates. Statistical analyses should account for the experimental design and avoid treating every cell as an independent sample. Pseudobulk approaches, which aggregate counts within cell types for each biological sample, are often useful for differential-expression testing across biological replicates.
  • Single-cell sequencing can also be used to study developmental processes and cell differentiation. During development, stem cells may gradually acquire specialized molecular characteristics and become distinct cell types. Single-cell transcriptomic data can help reconstruct these changes by identifying cells with related expression profiles and examining patterns of gene regulation. Trajectory inference methods attempt to arrange cells along developmental or differentiation paths. Tools such as Monocle, Slingshot, and related approaches can infer potential trajectories from single-cell data. RNA velocity methods use information from unspliced and spliced RNA to estimate aspects of the direction of transcriptional change. These methods provide computational hypotheses about cellular transitions rather than direct observations of lineage relationships or developmental history.
  • Single-cell sequencing is particularly valuable in cancer research because tumours often contain multiple genetically and functionally distinct cell populations. A tumour may include malignant cells, immune cells, fibroblasts, endothelial cells, and other components of the tumour microenvironment. Single-cell analysis can help identify tumour subpopulations, investigate mechanisms of treatment resistance, and examine interactions between malignant and non-malignant cells. Single-cell DNA sequencing can reveal genetic differences among tumour cells, while scRNA-seq can identify differences in gene expression and cellular states. These approaches may help investigate tumour evolution, clonal diversity, metastasis, and responses to therapy.
  • Single-cell DNA sequencing investigates genomic variation at the level of individual cells. It can be used to study somatic mutations, copy-number changes, chromosome abnormalities, and genetic heterogeneity within tissues or tumours. Unlike bulk DNA sequencing, which combines genetic material from many cells, single-cell DNA sequencing can help determine which variants occur together within particular cells. This information is valuable for reconstructing clonal relationships and studying how cell populations evolve. However, single-cell DNA sequencing can be technically challenging because each cell contains a limited amount of DNA. Whole-genome amplification may be required, and amplification bias, allelic dropout, uneven coverage, and technical errors can complicate variant detection.
  • Single-cell chromatin accessibility sequencing investigates which regions of DNA are accessible to regulatory proteins and transcription factors. Single-cell ATAC sequencing (scATAC-seq) uses the Tn5 transposase enzyme to insert sequencing adapters into accessible chromatin regions. Sequencing identifies genomic regions that were accessible in individual cells or cellular compartments. This information can help identify candidate promoters, enhancers, and other regulatory elements involved in gene expression. scATAC-seq data are usually analysed using specialized tools that process fragment files, construct accessibility matrices, identify peaks, and compare chromatin accessibility among cell populations.
  • Single-cell epigenomic technologies also investigate DNA methylation and other regulatory characteristics. DNA methylation is an epigenetic modification that can influence gene regulation and cellular identity. Single-cell methylation sequencing can reveal differences in methylation patterns among individual cells. Other approaches investigate histone modifications and chromatin-associated proteins. These methods provide information about gene regulation that is not directly available from RNA sequencing alone. However, single-cell epigenomic assays may have sparse coverage, complex experimental requirements, and specialized computational workflows.
  • Multi-omics approaches combine two or more molecular measurements from the same cell or from corresponding cell populations. Single-cell multi-omics can combine gene expression with chromatin accessibility, protein abundance, DNA methylation, or other molecular features. For example, simultaneous RNA and ATAC profiling can help connect gene-expression patterns with accessible regulatory regions. CITE-seq combines RNA sequencing with measurements of selected cell-surface proteins using antibody-derived tags. Other methods measure immune-cell receptor sequences together with gene expression, allowing researchers to connect cellular identity with T-cell or B-cell receptor information. Multi-omics methods provide a broader view of cellular biology but require careful normalization, integration, and interpretation of data from different measurement technologies.
  • Spatial transcriptomics is closely related to single-cell sequencing but provides an additional dimension: the location of molecular measurements within tissue. Conventional single-cell sequencing usually separates cells from their original tissue architecture, which can remove information about their physical positions and neighbouring cells. Spatial transcriptomics methods preserve spatial context by measuring gene expression at defined tissue locations. Some approaches provide measurements at approximately cellular or subcellular resolution, while others measure multiple cells within a single capture location. Certain spatial technologies can combine sequencing-based measurements with imaging or targeted molecular detection. Spatial data can be integrated with single-cell reference datasets to help identify cell types and understand tissue organization. Spatial transcriptomics is therefore a complementary field rather than a synonym for single-cell sequencing.
  • Immune-cell profiling is another important application. Single-cell sequencing can investigate the diversity and activation states of T cells, B cells, natural killer cells, and other immune populations. Single-cell immune-receptor sequencing can identify T-cell receptor (TCR) and B-cell receptor (BCR) sequences, helping researchers study clonal expansion, immune diversity, and antigen-associated responses. When receptor information is combined with gene-expression profiles, it becomes possible to examine the relationship between receptor identity and cellular function. These methods are useful in immunology, vaccine research, autoimmune disease studies, infectious disease research, and cancer immunotherapy.
  • Single-cell sequencing also has important applications in neuroscience. The brain contains many neuronal and non-neuronal cell types, each with distinct molecular characteristics. Single-cell and single-nucleus RNA sequencing can help classify neuronal subtypes, identify glial populations, and investigate molecular changes associated with development, aging, and neurological disorders. Single-cell methods can also be used to study cellular responses to injury, inflammation, and disease. Because neurons can be difficult to isolate intact and adult brain tissue may be challenging to dissociate, single-nucleus sequencing is often particularly useful in neuroscience research.
  • In developmental biology and stem-cell research, single-cell sequencing helps investigate how cells acquire specialized functions and how tissues form. Researchers can identify intermediate cell states, examine gene-regulatory programmes, and study variation among cells during differentiation. In regenerative medicine, these approaches can help characterize stem-cell populations, assess differentiation protocols, and identify unwanted or unexpected cell types. However, computationally inferred developmental trajectories do not by themselves prove that one observed cell directly developed into another. Lineage-tracing experiments and other experimental evidence may be needed to establish cellular ancestry.
  • Single-cell sequencing has several technical limitations. The amount of genetic material or RNA in an individual cell is small, which can lead to incomplete molecular capture and amplification bias. Some transcripts may not be detected even when they are present, resulting in what is commonly called a dropout event. Low-quality cells, doublets, ambient RNA, barcode errors, and batch effects can influence the data. Ambient RNA refers to extracellular RNA released from damaged or lysed cells, which may be mistakenly assigned to other cells during library preparation. Computational methods can estimate and reduce some of these effects, but they cannot fully recover all missing biological information.
  • Batch effects occur when technical differences between experimental runs, sample preparation procedures, sequencing batches, or laboratories influence measured expression profiles. These effects can create artificial differences between samples or obscure genuine biological variation. Integration methods are used to align datasets and reduce unwanted technical variation. Tools such as Seurat, Harmony, and scVI support different approaches to data integration and representation learning. Integration should be performed carefully because excessive correction may remove genuine biological differences, particularly when the samples represent different disease states, treatments, or developmental conditions.
  • The interpretation of single-cell sequencing data also requires attention to biological replication. A large number of cells from one individual does not necessarily provide the same statistical information as samples collected from many independent individuals. For studies comparing disease and control groups, researchers should include appropriate numbers of biological replicates and use statistical methods that reflect the sample-level experimental design. Cell-level observations can provide detailed information about cellular heterogeneity, but conclusions about population-level biological differences require suitable replication and statistical analysis.
  • Reproducibility is essential in single-cell sequencing research. Researchers should document sample collection, tissue processing, cell isolation, viability, library preparation chemistry, sequencing platform, reference genome, gene annotation version, barcode structure, and software versions. Original sequencing data, processed count matrices, cell metadata, quality-control reports, analysis scripts, and parameter settings should be preserved. Public repositories such as the Gene Expression Omnibus (GEO), ArrayExpress resources, and Single Cell Expression Atlas can support data sharing and reuse, subject to applicable privacy and consent requirements. Workflow managers such as Snakemake and Nextflow, container technologies such as Docker and Apptainer, and version control systems such as Git can support reproducible computational analysis.
  • Single-cell sequencing is transforming biological research by providing detailed information about cellular diversity and molecular variation. It enables researchers to identify rare cell populations, characterize cell states, investigate developmental processes, study tumour heterogeneity, and examine interactions between different cellular populations. Its applications extend across genomics, transcriptomics, epigenomics, immunology, neuroscience, developmental biology, and precision medicine. As sequencing technologies, molecular barcoding systems, spatial methods, and computational tools continue to develop, single-cell approaches are becoming increasingly important for understanding complex biological systems. Reliable results depend on suitable experimental design, high-quality sample preparation, appropriate sequencing methods, rigorous quality control, and careful bioinformatics interpretation.
Author:

Leave a Reply

Your email address will not be published. Required fields are marked *