Single-Cell Sequencing Analysis Pipeline: From Raw Data to Biological Insights

Single-cell RNA sequencing (scRNA-seq) analysis requires a structured computational workflow that converts raw sequencing data into interpretable biological findings. This article provides an end-to-end pipeline overview covering alignment, quality control, normalization, clustering, and annotation, with tool recommendations and troubleshooting guidance for students, researchers, analysts, and life-science professionals. The workflow described here applies to standard scRNA-seq datasets generated by droplet-based and plate-based platforms, with attention to the decisions that most influence downstream results.

Scope and Reader Context

Single-cell sequencing has enabled gene expression measurement at the resolution of individual cells, creating opportunities to resolve cellular heterogeneity that bulk RNA-seq cannot capture. The analysis of these datasets demands specialized computational tools because of the high data complexity and the absence of universal standardization across the field. This article serves as a practical reference for bioinformaticians who need to build, evaluate, or troubleshoot a single-cell analysis pipeline. The content assumes familiarity with basic RNA-seq concepts but does not require prior single-cell analysis experience.

The pipeline described here covers the canonical analytical workflow: read mapping, quality control, gene expression quantification, normalization, feature selection, dimensionality reduction, cell clustering, and downstream analyses including trajectory inference and differential expression. Each stage is presented with concrete decisions, tool options, and common failure patterns so that analysts can apply the guidance directly to their own datasets.

At a Glance: Pipeline Stage Decision Table

The table below summarizes the main pipeline stages, the key decisions at each stage, and the primary failure modes that analysts encounter. Use this table as a quick reference before reading the detailed sections that follow.

Data Inputs and Experimental Design Considerations

The analysis pipeline begins before sequencing data are generated. Experimental design decisions influence every downstream computational step, and analysts should review these choices with the research team before processing begins.

Sample Preparation and Cell Preservation

Single-cell RNA-seq typically requires fresh, high-quality single-cell suspensions processed immediately to preserve transcriptional profiles. This constraint complicates sample collection at remote sites and creates difficulties for studies with long preparation times. Commercial assays now enable cell preservation at the point of collection through fixation or cryopreservation, allowing processing to occur months later. A multisite assessment of three preservation platforms found that fixed or cryopreserved samples can be processed successfully after extended storage, though performance varies across platforms and should be validated for each experimental context.

When samples cannot be processed immediately, analysts should document the preservation method, storage duration, and any deviations from the manufacturer protocol. These records become essential when interpreting quality metrics and comparing results across batches.

Sequencing Platform Selection

The choice of single-cell platform affects data characteristics and analysis requirements. Droplet-based ultra-high-throughput systems enable profiling of thousands to millions of cells per experiment, while plate-based methods offer different tradeoffs in throughput and sensitivity. Comparative evaluations of droplet-based systems have documented differences in performance that influence downstream analysis choices.

Analysts should record the platform, chemistry version, and sequencing configuration for every dataset. This metadata supports informed decisions about alignment parameters, quality thresholds, and integration strategies when combining data from multiple sources.

Public Data Repositories and Data Sharing

Many research questions can be addressed using publicly available single-cell datasets, which reduces the cost barrier associated with data generation. The National Center for Biotechnology Information provides access to a wide range of genomic data resources, including single-cell sequencing datasets deposited by the research community. The EMBL-EBI Training program offers educational materials on accessing and analyzing these public resources.

When using public data, analysts must verify the data use restrictions and attribution requirements associated with each dataset. The NIH Genomic Data Sharing Policy establishes expectations for responsible data sharing and use, and analysts should confirm that their intended analyses comply with applicable policies.

Read Alignment and Quantification

The first computational stage transforms raw sequencing reads into a count matrix that records gene expression levels for each cell. This stage requires careful attention to reference genome selection, alignment parameters, and quality assessment.

Reference Genome and Annotation Selection

The choice of reference genome and gene annotation directly affects quantification accuracy. Analysts should use the same reference genome version and annotation source for all samples within a study to ensure comparability. When combining data from multiple studies, differences in reference versions can introduce artifacts that mimic biological variation.

For human and mouse data, the standard reference genomes are widely available through public repositories. For non-model organisms, analysts may need to construct custom references from genome assemblies and annotation files. The quality of the reference annotation determines the reliability of gene-level quantification, and poorly annotated genomes produce noisy count matrices that complicate downstream analysis.

Alignment Strategies

Single-cell alignment differs from bulk RNA-seq in several respects. The high number of cells and the presence of cell barcodes and unique molecular identifiers require specialized handling during alignment and quantification. Several alignment tools are available, and systematic evaluations have compared their performance across different library preparation protocols.

A systematic evaluation of single-cell RNA-seq analysis pipelines found that the choice of mapping approach influences downstream results, though normalization and library preparation protocols had the largest impact on analysis outcomes. Analysts should select an alignment strategy that is compatible with their platform and that has been validated for their data type.

For immune-focused studies, standard alignment pipelines may miss information that a supplemental alignment approach can capture. An immune-focused supplemental alignment pipeline was shown to recover allele-specific MHC-I regulation information that dominant single-cell RNA-seq analyses missed. Analysts studying immune cell populations should consider whether their alignment strategy captures the full range of relevant biological signals.

Quantification and Count Matrix Generation

The output of alignment and quantification is a count matrix with cells as columns and genes as rows. Each entry records the number of unique molecular identifiers detected for a given gene in a given cell. This matrix serves as the input for all downstream analysis stages.

Quality assessment at this stage includes examining the total number of reads, the fraction of reads mapped to the reference genome, the fraction of reads assigned to genes, and the number of genes detected per cell. These metrics provide an early indication of data quality and can identify problems with sample preparation or sequencing.

Quality Control and Cell Filtering

Quality control removes low-quality cells and genes that would otherwise introduce noise into downstream analyses. This stage requires balancing sensitivity against specificity, because overly aggressive filtering can remove real cell populations while insufficient filtering leaves technical artifacts.

Cell-Level Quality Metrics

Standard quality control metrics for single-cell data include the number of genes detected per cell, the total number of unique molecular identifiers per cell, and the fraction of reads mapped to mitochondrial genes. Cells with very low gene counts may represent empty droplets or broken cells, while cells with very high mitochondrial fractions often indicate stressed or dying cells.

The scReady pipeline integrates essential quality control steps including ambient RNA removal, doublet detection, and cell and gene filtering based on mitochondrial content and customizable thresholds. This automated approach standardizes preprocessing and produces diagnostic plots and quality control reports that support reproducible analysis.

Doublet Detection

Doublets occur when two cells are captured in the same droplet or well and sequenced together. Doublets can be mistaken for novel cell types or intermediate states, so their detection and removal is an important quality control step. Multiple computational methods are available for doublet detection, and the choice of method can affect downstream results.

Analysts should record the expected doublet rate for their platform and compare the detected doublet fraction against this expectation. Unusually high doublet rates may indicate problems with cell loading or droplet generation.

Ambient RNA Contamination

Ambient RNA refers to free-floating RNA molecules in the cell suspension that are captured and sequenced alongside cellular RNA. This contamination can obscure true gene expression patterns, particularly for highly expressed genes. Computational methods for ambient RNA removal estimate the contamination fraction and correct the count matrix accordingly.

The decision to apply ambient RNA removal should be based on the expected contamination level for the platform and protocol. Some platforms have higher ambient RNA levels than others, and the choice of correction method can influence downstream clustering and differential expression results.

Filtering Threshold Selection

The selection of filtering thresholds requires examination of the data distribution instead of blind application of default values. Analysts should plot the distribution of quality metrics and identify natural breakpoints that separate high-quality cells from low-quality cells.

The scReady pipeline allows customizable thresholds for cell and gene filtering, enabling analysts to adapt the filtering strategy to their specific dataset. This flexibility is important because optimal thresholds vary across tissues, platforms, and experimental conditions.

Normalization and Batch Correction

Normalization adjusts for technical variation in sequencing depth and capture efficiency so that biological differences can be detected reliably. Batch correction addresses systematic differences between samples processed at different times or under different conditions.

Normalization Methods

A systematic evaluation of single-cell RNA-seq analysis pipelines found that the choice of normalization method has the largest impact on analysis outcomes, particularly for detecting asymmetric differential expression. The study evaluated seven normalization approaches and found that normalization dominates pipeline performance in asymmetric differential expression setups.

The same evaluation found that library preparation protocols determine the ability to detect symmetric expression differences. This interaction between experimental and computational choices means that analysts should consider their library preparation method when selecting a normalization approach.

Integration and Batch Effect Correction

When combining data from multiple samples or studies, batch effects can obscure biological variation or create spurious clusters. Integration methods align cells across batches while preserving biological differences. The choice of integration method and its parameters requires careful evaluation to avoid overcorrection.

The IMMUcan database developed a data processing pipeline organized in four steps: data collection, data processing with quality control and sample integration, supervised cell annotation with a cell ontology classifier, and an interface for analyzing the tumor microenvironment. This framework demonstrates how integration can be combined with annotation to support meta-analysis across datasets.

Multi-Species Considerations

For experiments that profile multiple species simultaneously, normalization and downstream analysis require modifications compared to single-species pipelines. Multi-species transcriptomics must account for the relative abundances of each organism within the sample, which often requires enrichment methods and yields differences in total read counts across samples. Best practices for multi-species differential expression analysis describe modifications to alignment, quantification, and downstream analysis steps.

Analysts working with multi-species data should verify that their normalization approach accounts for the different library compositions and that differential expression results are interpreted in light of the relative abundances of each species.

Feature Selection and Dimensionality Reduction

Feature selection identifies the genes that carry the most biological information, and dimensionality reduction projects the high-dimensional expression space into a lower-dimensional representation that supports visualization and clustering.

Highly Variable Gene Selection

The selection of highly variable genes reduces the dimensionality of the dataset and focuses the analysis on genes that vary across cells. The number of selected genes and the method used for selection can influence clustering results and the resolution of rare cell populations.

Current best practices for single-cell RNA-seq analysis recommend selecting a set of variably expressed genes for downstream analysis. The tutorial by the authors of the best practices review integrates these recommendations into a workflow applied to a public dataset, providing a documented case study for new entrants to the field.

Principal Component Analysis

Principal component analysis is the most commonly used dimensionality reduction method for single-cell data. The number of principal components retained for downstream analysis affects the balance between capturing biological signal and including technical noise. Analysts should examine the variance explained by each component and select a cutoff that retains meaningful biological variation.

The choice of the number of principal components is dataset-specific and should be evaluated for each analysis. Retaining too few components can lose rare cell populations, while retaining too many can introduce noise that obscures clustering structure.

Visualization Embeddings

Uniform manifold approximation and projection and t-distributed stochastic neighbor embedding are commonly used for visualizing single-cell data in two dimensions. These methods have parameters that affect the visual separation of cell populations, and the choice of parameters can influence interpretation.

Visualization embeddings should be treated as exploratory tools instead of definitive representations of cell relationships. The scFlowVis system was designed to support exploration, comparison, and workflow tracking across the single-cell RNA-seq pipeline, addressing the need for visual analytics tools that support the complete workflow instead of isolated stages.

Clustering and Cell Type Identification

Clustering groups cells with similar expression profiles, and cell type identification assigns biological labels to the resulting clusters. This stage requires careful attention to resolution parameters and validation of cluster assignments.

Clustering Resolution

The clustering resolution parameter determines the number of clusters identified. Low resolution produces fewer, larger clusters that may merge distinct cell types, while high resolution produces more clusters that may split a single cell type into multiple subgroups.

A computational workflow for automatic extraction of mesenchymal stromal cells from dental pulp single-cell data found that the choice of clustering resolution is determined empirically, potentially leading to under- or over-segmentation of target cell populations. The workflow systematically evaluated marker set performance and demonstrated that a six-gene marker set provided higher cluster separation than a three-gene basic marker set.

Marker Gene-Based Annotation

Cell type annotation typically relies on marker genes that are known to be expressed in specific cell types. Analysts can annotate clusters manually by examining marker gene expression, or they can use automated annotation tools that compare cluster expression profiles against reference datasets.

The MetaNeighbor framework quantifies the degree to which cell types replicate across datasets and enables rapid identification of clusters with high similarity. Application of this framework to neuronal subtypes found that 24 of 45 subtypes had evidence of replication, enabling the identification of robust candidate marker genes.

Reference-Based Annotation

Reference-based annotation assigns cell type labels by comparing each cell or cluster against a reference dataset with known cell type annotations. This approach can improve consistency across datasets and reduce the subjectivity of manual annotation.

The IMMUcan database provides supervised cell annotation with a cell ontology classifier of the tumor microenvironment, enabling consistent annotation across cancer types. This resource encompasses 144 datasets on 56 different cancer types with detailed clinical, technological, and biological information.

Validation of Cell Type Assignments

Cell type assignments should be validated using multiple lines of evidence. Analysts should examine the expression of canonical marker genes, compare cluster expression profiles against reference data, and assess whether the identified cell types are biologically plausible for the tissue under study.

The replicability of cell types defined by single-cell RNA-seq data can be characterized using methods like MetaNeighbor, which measures cell-type replication across datasets. This approach helps identify which cell types are robustly defined and which may be artifacts of a particular dataset or analysis pipeline.

Downstream Analyses and Biological Interpretation

After clustering and annotation, analysts can perform a range of downstream analyses to address specific biological questions. These analyses include differential expression, trajectory inference, and integration with other data types.

Differential Expression Analysis

Differential expression analysis identifies genes that are expressed at different levels between cell types or conditions. The choice of differential expression method can influence results, and systematic evaluations have compared the performance of different approaches.

The systematic evaluation of single-cell RNA-seq analysis pipelines found that the choice of differential expression testing approach interacts with normalization and library preparation choices. Analysts should select a differential expression method that is appropriate for their data structure and that has been validated for single-cell data.

Trajectory Inference

Trajectory inference reconstructs developmental or activation trajectories from single-cell data by ordering cells along a continuous path. This analysis can reveal intermediate cell states and the relationships between cell populations.

Trajectory inference methods make different assumptions about the underlying biology, and the choice of method can influence the inferred trajectory structure. Analysts should evaluate multiple methods and assess whether the inferred trajectories are consistent with known biology.

Integration with Spatial Transcriptomics

Spatial transcriptomics adds spatial information to single-cell gene expression data, enabling the mapping of cell types to their locations within tissues. The ssREAD database provides an analysis suite for cell clustering, identification of differentially expressed and spatially variable genes, cell-type-specific marker genes and regulons, and spot deconvolution for integrative analysis.

This database encompasses 1,053 samples from 67 Alzheimer's disease-related single-cell and single-nucleus RNA-seq studies, totaling over 7 million cells, along with 381 spatial transcriptomics datasets. The integration of single-cell and spatial data can provide insights into cell-cell interactions and tissue organization.

Signature Development and Risk Prediction

Single-cell data can be used to develop gene signatures that predict clinical outcomes or disease progression. An integrative analysis pipeline developed a neuroendocrine cell-intrinsic gene signature from a comprehensive single-cell reference atlas of prostate cancer, identifying 762 high-quality neuroendocrine markers and developing an R package to predict neuroendocrine prostate cancer risk scores.

The signature demonstrated predictive capabilities for prognosis and therapy responsiveness, illustrating how single-cell analysis can generate clinically relevant tools. Analysts developing signatures should validate their results across multiple independent datasets to ensure robustness.

Reproducibility and Workflow Management

Reproducibility is a central concern in single-cell analysis because the many computational choices can produce different results from the same input data. Workflow management tools and documentation practices support reproducible analysis.

Containerized Pipelines

Containerized pipelines package analysis software and dependencies into portable environments that can be run consistently across different computing systems. The scReady pipeline is designed for both single machines and high-performance computing clusters, ensuring flexibility, scalability, and reproducibility.

The SC-Framework provides a FAIR-compliant, layered, semi-interactive analysis environment that combines a Python package with a structured series of Jupyter Notebooks. Containerized releases support long-term reproducibility on both local machines and high-performance clusters, and benchmarking demonstrates scalability to nearly one million cells.

FAIR Data Principles

The FAIR Guiding Principles describe the characteristics that data should have to support findability, accessibility, interoperability, and reusability. Applying these principles to single-cell data and analysis workflows supports scientific reproducibility and data sharing.

The SC-Framework ensures traceability and findability via self-documenting data objects and configuration-defined directory structures. This approach addresses the gap between rigid automated pipelines and flexible but unstructured toolkit-based approaches.

Documentation and Record Keeping

Analysts should document all computational choices, including software versions, parameter settings, and reference genome versions. This documentation enables others to reproduce the analysis and supports interpretation of results.

The scReady pipeline generates diagnostic plots and a comprehensive quality control report, providing a record of the preprocessing decisions and their outcomes. Similar documentation practices should be applied to all pipeline stages.

Common Failure Patterns and Troubleshooting

Understanding common failure patterns helps analysts diagnose problems quickly and avoid costly reanalysis. The following patterns are frequently encountered in single-cell analysis.

Low Mapping Rates

Low mapping rates can result from reference genome mismatches, contamination, or problems with library preparation. Analysts should examine the distribution of mapping rates across cells and investigate the cause of low rates before proceeding.

Excessive Mitochondrial Reads

High mitochondrial fractions indicate stressed or dying cells. While some mitochondrial reads are expected, excessive fractions suggest problems with cell isolation or sample processing. The filtering threshold for mitochondrial content should be based on the tissue and protocol.

Batch Effects Obscuring Biology

When combining data from multiple samples, batch effects can create clusters that reflect technical variation instead of biological differences. Integration methods can correct for batch effects, but overcorrection can remove genuine biological variation. Analysts should evaluate integration results by examining whether known biological differences are preserved.

Overclustering and Underclustering

The choice of clustering resolution can produce too many or too few clusters. Overclustering splits real cell types into multiple subgroups, while underclustering merges distinct cell types. Analysts should evaluate clustering results at multiple resolutions and use marker gene expression to assess whether clusters correspond to known cell types.

Inconsistent Cell Type Annotation

Cell type annotation can vary across analysts and across datasets, hampering meta-analysis. The use of standardized naming conventions and reference-based annotation tools can improve consistency. The MetaNeighbor framework addresses this challenge by quantifying cell-type replication across datasets.

Limitations and Interpretation Boundaries

Single-cell analysis has inherent limitations that affect the interpretation of results. Analysts should communicate these limitations clearly when reporting findings.

Technical Noise and Dropout

Single-cell data contain high levels of technical noise, including dropout events where genes are not detected despite being expressed. This noise can obscure true biological variation and complicate the interpretation of expression levels.

Resolution Limits

The resolution of single-cell analysis is limited by sequencing depth and the sensitivity of the platform. Rare cell populations may not be detected, and subtle differences between cell states may not be resolved.

Computational Method Dependence

Results can vary depending on the computational methods chosen at each pipeline stage. The systematic evaluation of single-cell RNA-seq analysis pipelines found that choices of normalization and library preparation protocols have the biggest impact on analysis outcomes. Analysts should acknowledge this method dependence when interpreting and reporting results.

Cross-Study Comparability

Comparing results across studies is complicated by differences in experimental protocols, computational pipelines, and cell type naming conventions. The IMMUcan database addresses this challenge by providing a fully integrated single-cell database with standardized processing and annotation.

Safety and Regulatory Context

Single-cell analysis involving human data is subject to ethical and regulatory requirements. Analysts should ensure that their use of human data complies with applicable policies and that patient privacy is protected.

The NIH Genomic Data Sharing Policy establishes expectations for the responsible sharing and use of genomic data generated through NIH-funded research. Analysts using human data should review the terms of data use agreements and ensure that their analyses comply with the approved uses.

For clinical applications, single-cell signatures and biomarkers require validation before use in patient care. The neuroendocrine prostate cancer signature developed from single-cell data demonstrated predictive capabilities in research settings, but clinical implementation requires additional validation and regulatory approval.

Professional Escalation Criteria

Analysts should escalate concerns to appropriate professionals when they encounter situations beyond their expertise or when results have implications for patient care or regulatory compliance.

Data Quality Concerns

If quality control metrics indicate systematic problems with a dataset, analysts should consult with the laboratory that generated the data to investigate potential causes. Persistent quality issues may require repeating the experiment or adjusting the experimental protocol.

Clinical Interpretation

When single-cell analysis results are used to inform clinical decisions, interpretation should involve professionals with appropriate clinical expertise. Analysts should not make clinical recommendations based solely on computational results without appropriate clinical consultation.

Regulatory Compliance

If analysis involves human data subject to regulatory oversight, analysts should consult with institutional review boards or data governance professionals to ensure compliance with applicable policies.

Frequently Asked Questions

What is the first step in a single-cell RNA-seq analysis pipeline?

The first step is read alignment and quantification, where raw sequencing reads are mapped to a reference genome and gene expression levels are quantified for each cell. This produces a count matrix that serves as the input for all downstream analysis stages. The choice of reference genome version and alignment tool affects the quality of the count matrix and should be documented carefully.

How do I choose quality control thresholds for cell filtering?

Quality control thresholds should be based on the distribution of quality metrics in your specific dataset instead of default values. Examine the distribution of gene counts, unique molecular identifier counts, and mitochondrial fractions, and identify natural breakpoints that separate high-quality cells from low-quality cells. The scReady pipeline allows customizable thresholds and generates diagnostic plots to support this decision.

What is the difference between normalization and batch correction?

Normalization adjusts for technical variation in sequencing depth and capture efficiency within a sample, while batch correction addresses systematic differences between samples processed under different conditions. Both steps are important for detecting biological differences, and the choice of methods can significantly influence downstream results.

How many clusters should I expect from clustering analysis?

The number of clusters depends on the biological complexity of the tissue and the clustering resolution parameter. There is no universal answer, and analysts should evaluate clustering results at multiple resolutions and validate cluster assignments using marker gene expression. The choice of clustering resolution is determined empirically and can lead to under- or over-segmentation if not carefully evaluated.

What is the best way to annotate cell types?

Cell type annotation can be performed manually using marker gene expression or automatically using reference-based tools. Manual annotation requires biological knowledge of the tissue under study, while reference-based annotation improves consistency across datasets. The MetaNeighbor framework can quantify the degree to which cell types replicate across datasets and support the identification of robust marker genes.

How can I make my single-cell analysis reproducible?

Reproducibility requires documenting all computational choices, including software versions, parameter settings, and reference genome versions. Containerized pipelines like scReady and the SC-Framework package analysis software into portable environments that can be run consistently across computing systems. The FAIR Guiding Principles provide a framework for making data and workflows findable, accessible, interoperable, and reusable.

Can I use public single-cell datasets for my research?

Public single-cell datasets are available through repositories such as the National Center for Biotechnology Information and can be used to address focused research questions without generating new data. When using public data, verify the data use restrictions and attribution requirements associated with each dataset, and ensure that your intended analyses comply with applicable policies.

What should I do if my clustering results do not match known biology?

If clustering results do not match known biology, first check whether quality control filtering was appropriate and whether batch effects were adequately corrected. Then evaluate whether the clustering resolution is appropriate for the biological question. If results remain inconsistent with known biology, consider whether the reference genome annotation is complete and whether the platform has known limitations for the tissue under study.

Related Bioinformatics Guides

• Single-Cell RNA-seq Clustering and Cell-Type Annotation Pipelines
• Single-Cell RNA Sequencing: From Bulk to Resolution
• Master Guide: Single-Cell RNA Sequencing Bioinformatics Workflows
• Single-Cell RNA-Seq Analysis Pipelines for Veterinary Immunology
• Single-Cell RNA-Seq Normalization: Batch Effect Correction and Dimension Reduction (PCA, t-SNE, UMAP)

References and Further Reading

• EMBL-EBI Training. European Bioinformatics Institute.
• NCBI Data Resources. National Center for Biotechnology Information.
• Genomic Data Sharing Policy. National Institutes of Health.
• The FAIR Guiding Principles. Scientific Data.
• Current best practices in single-cell RNA-seq analysis: a tutorial.. Molecular systems biology, 2019.
• Single-Cell RNA Sequencing Analysis: A Step-by-Step Overview.. Methods in molecular biology (Clifton, N.J.), 2021.
• Single-Cell Map of Diverse Immune Phenotypes in the Breast Tumor Microenvironment.. Cell, 2018.
• A single-cell and spatial RNA-seq database for Alzheimer's disease (ssREAD).. Nature communications, 2024.
• Integrated analysis of single-cell and bulk transcriptomics develops a robust neuroendocrine cell-intrinsic signature to predict prostate cancer progression.. Theranostics, 2024.
• Single-Cell RNA-Seq Technologies and Computational Analysis Tools: Application in Cancer Research.. Methods in molecular biology (Clifton, N.J.), 2022.
• Best practices on the differential expression analysis of multi-species RNA-seq.. Genome biology, 2021.
• Meta-Analysis of Human Cancer Single-Cell RNA-Seq Datasets Using the IMMUcan Database.. Cancer research, 2023.
• Automatic extraction of mesenchymal stromal cells from single-cell RNA-sequencing data of human dental pulp.. 2026.
• Designing scFlowVis: Visual Analytics for Single-Cell RNA Sequencing Analysis.. 2026.
• SC-framework: A robust and FAIR semi-interactive environment for single-cell resolution datasets.. 2026.
• Multisite Assessment of Methods for Cell Preservation Upstream of Single-Cell RNA Sequencing.. 2026.
• A systematic evaluation of single cell RNA-seq analysis pipelines. Nature Communications, 2019.
• scReady - an automated and accessible pipeline for single-cell RNA-Seq preprocessing: Empowering novice bioinformaticians. Wellcome Open Research, 2026.
• Supporting Single Cell RNA-seq Analysis at Harvard - A Community Approach.. Journal of Biomolecular Techniques, 2019.
• Tutorial: guidelines for the computational analysis of single-cell RNA sequencing data. Nature Protocols, 2020.
• Characterizing the replicability of cell types defined by single cell RNA-sequencing data using MetaNeighbor. Nature Communications, 2018.
• SINCERA: A Pipeline for Single-Cell RNA-Seq Profiling Analysis. Plos Computational Biology, 2015.
• scDown: A Pipeline for Single-Cell RNA-Seq Downstream Analysis. International Journal of Molecular Sciences, 2025.
• SC1: A web-based single cell RNA-seq analysis pipeline. IEEE International Conference on Computational Advances in Bio and Medical Sciences Iccabs, 2018.
• Comparative Analysis of Droplet-Based Ultra-High-Throughput Single-Cell RNA-Seq Systems. Molecular Cell, 2019.
• An immune-focused supplemental alignment pipeline captures information missed from dominant single-cell RNA-seq analyses, including allele-specific MHC-I regulation. Frontiers in Immunology, 2025.

This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.