Single-Cell Sequencing Services: How to Choose a Provider

Single-cell sequencing services have become a standard outsourcing option for research groups that need cell-level resolution but lack in-house library preparation, sequencing capacity, or bioinformatics expertise. Choosing a provider requires evaluating platform compatibility, data quality controls, turnaround time, cost structure, and the provider's ability to deliver analysis-ready data instead of raw files alone. This article gives researchers, students, and analysts a practical framework for comparing single-cell sequencing service providers and for asking the questions that separate reliable vendors from those that will generate unusable or poorly documented data.

At a Glance

The table below summarizes the main decision points when evaluating a single-cell sequencing service provider. Use it as a starting checklist before you request quotes or send samples.

Understanding What Single-Cell Sequencing Services Actually Deliver

A single-cell sequencing service provider handles the workflow from cell suspension to count matrix. The core deliverable is a table of gene expression values for each individual cell, along with metadata about cell barcodes and quality metrics. Some providers also offer clustering, cell type annotation, differential expression analysis, and trajectory inference as paid add-ons.

The most common service is single-cell RNA sequencing (scRNA-seq), which measures the transcriptome of individual cells. Related services include single-nucleus RNA sequencing for frozen or hard-to-dissociate tissues, and assays that capture additional modalities such as cell surface proteins or chromatin accessibility. The choice of assay depends on your biological question and your sample type.

Single-cell approaches have been used to build reference atlases of complex tissues, including the human lung endothelium, where reanalysis of multiple datasets identified previously indistinguishable endothelial subpopulations [12]. Similar studies have profiled the aging Drosophila brain and identified nearly all transcriptional states across the lifespan [10]. These examples show that the value of single-cell data depends heavily on the quality of the underlying libraries and the care taken during data processing.

Core Principles for Evaluating Providers

Platform Options and Their Tradeoffs

Providers typically offer one or more commercial platforms. The most widely used is the 10x Genomics droplet-based system, which captures individual cells in nanoliter-scale gel beads. This platform requires a fresh or properly preserved single-cell suspension and is the default choice for many service providers.

Alternative platforms use combinatorial barcoding, where cells are split and pooled across multi-well plates. SPLiT-seq is one such approach that applies unique barcode combinations through sequential splitting and pooling steps [18]. These platforms can be more cost-effective for large numbers of cells and may accommodate fixed samples that cannot be processed immediately.

A multisite study evaluated three preservation-compatible platforms: 10x Genomics FLEX, Parse Biosciences Evercode WT v2, and Honeycomb Bio HIVE [14]. The study used total leukocytes and peripheral blood mononuclear cells from a single donor, with a 21-color flow cytometry panel as a reference. Samples were fixed or cryopreserved according to each platform protocol, then processed in parallel by two technicians and distributed to multiple core facilities. The evaluation covered standard quality control metrics, gene and transcript detection sensitivity, cell type discovery, differential expression, and correlation with the flow cytometry reference [14].

The practical implication is that platform choice affects more than price. Gene detection sensitivity varies across platforms, and the preservation method you use at the point of collection may limit which providers can process your samples. Ask each provider which platforms they support and whether they have validated those platforms on your sample type.

Fresh, Fixed, and Cryopreserved Samples

The condition of your cells at the time of processing is the single largest determinant of data quality. Fresh single-cell suspensions processed immediately preserve transcriptional profiles most faithfully. However, this constraint complicates studies where samples are collected at remote sites or where preparation times are long [14].

Commercial preservation assays now allow fixation or cryopreservation at the point of collection, with processing occurring months later [14]. The multisite study found that performance varied across platforms, which means you cannot assume that a preservation method validated for one platform will work on another. If your samples cannot be processed immediately, ask the provider which preservation methods they have tested and request evidence from their own validation runs.

For tissues that are difficult to dissociate into single cells, single-nucleus RNA sequencing is an alternative. This approach works on frozen tissue and avoids the dissociation-induced gene expression changes that can occur with enzymatic digestion. Providers that offer both single-cell and single-nucleus services can advise on which approach suits your tissue type.

Data Quality Controls and Metrics

A reliable provider will report standard quality control metrics for every sample. These include the number of cells detected, the number of genes detected per cell, the total number of reads, the mapping rate to the reference genome, the fraction of reads mapping to mitochondrial genes, and the estimated doublet rate.

High mitochondrial read fractions typically indicate dying or stressed cells, because mitochondrial transcripts are retained when cytoplasmic mRNA is lost. Low gene counts per cell may indicate empty droplets or damaged cells. Doublet rates reflect the proportion of droplets that captured two or more cells, which can create artificial cell populations.

The provider should also report how these metrics were calculated and which thresholds were applied. Different count matrix generation pipelines produce different results, and the choice of pipeline affects downstream analysis [18]. A comparison of eight SPLiT-seq processing pipelines found that STARsolo, splitpipe, and alevin-fry splitp handled large datasets within reasonable time, while other pipelines were slow on large data [18]. The same study found that alevin-fry produced downstream results that were difficult to interpret, while STARsolo and splitpipe produced highly similar results [18].

Ask the provider which alignment and counting software they use, which reference genome version and annotation they align to, and whether they will share the exact commands or pipeline configuration. Reproducibility depends on this information being available.

Practical Workflow for Selecting a Provider

Step 1: Define Your Biological Question and Sample Constraints

Before contacting providers, write down your experimental design. Specify the tissue or cell type, the number of samples, the expected number of cells per sample, and whether you need to preserve cells at the collection site. If you are studying a rare cell population, you may need to sequence more cells to capture enough of that population. If you are comparing conditions, such as diseased versus control tissue, you need to ensure that all samples are processed with the same platform and chemistry.

Consider whether your study will benefit from integration with public datasets. Many single-cell studies deposit their data in public repositories, and reanalysis of existing data can provide reference atlases for your own cell type annotation [12]. The NCBI maintains a range of data resources that include single-cell expression data, and the EMBL-EBI offers training materials on data submission and retrieval [1][2]. If you plan to compare your data with public atlases, the provider should use a reference genome version and annotation that are compatible with those datasets.

Step 2: Request Detailed Quotes and Compare Line Items

Ask each provider for a written quote that separates library preparation, sequencing, and analysis costs. Sequencing depth is a major cost driver, and providers may quote different depths depending on the platform and the number of cells. For standard scRNA-seq, deeper sequencing per cell increases gene detection sensitivity up to a saturation point, beyond which additional reads add little information.

Ask whether the quote includes quality control steps such as cell counting, viability assessment, and library quantification. Some providers charge extra for these steps or for re-processing samples that fail quality thresholds. Ask what happens if a library fails: does the provider re-run the sample at no cost, or do you pay for a second attempt?

Analysis costs are often quoted separately. A basic deliverable may include raw FASTQ files and a count matrix. Additional analysis, such as clustering, cell type annotation, differential expression, or trajectory inference, may be billed per hour or per sample. Ask for a fixed price for a defined analysis scope so you can compare providers on an equal basis.

Step 3: Evaluate Bioinformatics Capabilities

The provider's bioinformatics team determines whether you receive usable data or a collection of files you cannot interpret. Ask which pipelines they use for alignment, counting, and quality control. Ask whether they provide processed objects in standard formats such as Seurat or Scanpy, and whether they include the code or configuration files used to generate those objects.

Ask how the provider handles batch effects when processing multiple samples. If you have samples collected at different times or processed in different batches, the provider should have a strategy for batch correction or at least for documenting batch structure in the metadata. The provider should also be able to advise on the number of cells needed per sample to detect the cell types or states you are interested in.

For studies that require regulatory network inference, tools such as SCENIC can reconstruct gene regulatory networks and identify cell states from scRNA-seq data [6]. SCENIC has been used to reveal regulatory heterogeneity linked to energy consumption in the aging Drosophila brain [10]. If your analysis plan includes this type of inference, ask whether the provider has experience running SCENIC or similar tools and whether they can deliver the required input formats.

Step 4: Assess Turnaround Time and Communication

Ask for a realistic timeline from sample receipt to data delivery. The timeline should include library preparation, sequencing, and primary analysis. Ask how the provider handles delays, such as instrument failures or reagent shortages, and whether they will communicate proactively if the timeline slips.

Ask who your point of contact will be and whether that person has hands-on experience with single-cell library preparation and analysis. A provider that assigns a dedicated project manager or bioinformatician to your project is more likely to catch quality issues early and to explain problems in terms you can act on.

Step 5: Check Data Storage, Transfer, and Sharing Policies

Genomic data may be subject to institutional, funder, or national data sharing policies. The NIH Genomic Data Sharing Policy sets expectations for data sharing and privacy protection for NIH-funded research [3]. If your project is funded by NIH or another agency with similar requirements, confirm that the provider can deliver data in formats that comply with those policies and that they can transfer data through secure channels.

Ask how long the provider will store your raw data and processed files. Some providers delete raw data after a set period unless you pay for extended storage. Ask whether you will receive all raw files, including intermediate files such as BAM alignments, or only the final count matrix. For reproducibility and for reanalysis with updated tools, you may need access to the raw FASTQ files.

Records and Measurements to Keep

Maintain a project record that documents every decision you make during provider selection and sample submission. This record should include the following items:

• The biological question and experimental design, including the number of samples and expected cell counts
• The sample collection and preservation protocol, including the exact reagents and timing used
• The provider quote, including all line items and any assumptions about sequencing depth or analysis scope
• The sample submission form and any quality control data the provider collected before library preparation
• The provider's quality control report for each sample, including cell counts, gene counts, mapping rates, and doublet rates
• The exact pipeline versions and parameters used for alignment and counting
• The dates of sample shipment, receipt, library preparation, sequencing, and data delivery
• Any communication with the provider about quality issues, delays, or re-runs

This record serves two purposes. First, it allows you to compare providers objectively if you need to repeat the experiment or scale up. Second, it provides the documentation needed for methods sections in papers and for data sharing submissions.

Common Failure Patterns and How to Avoid Them

Poor Cell Viability at Submission

The most common cause of failed single-cell experiments is low cell viability at the time of library preparation. Cells that are damaged or dying release ambient RNA that contaminates the capture droplets and inflates background signal. Providers should measure viability before library preparation and should reject samples below their stated threshold. Ask what that threshold is and what the provider does if your sample falls below it.

Incompatible Preservation Methods

If you preserve cells at the collection site, the preservation method must match the platform the provider uses. The multisite study of preservation methods found that performance varied across platforms, so a method that works for one provider may not work for another [14]. Confirm the preservation method with the provider before you collect samples, and ask for evidence that the provider has validated that method on your cell type.

Misaligned Sequencing Depth

Providers may quote a sequencing depth that is too low for your biological question. If you need to detect rare cell types or subtle gene expression differences, you may need more reads per cell than the provider's default. Ask the provider to explain their depth recommendation and to show how they determined it. If the provider cannot justify the depth, consider a different vendor.

Opaque Bioinformatics

Some providers deliver only raw FASTQ files and a count matrix, leaving alignment and quality control to you. This is acceptable if you have bioinformatics support, but it shifts the burden of quality assessment onto your team. If you need analysis-ready data, choose a provider that includes processed objects and a written analysis report. Ask to see an example report before you commit.

Batch Effects Across Samples

If your samples are processed in multiple batches, batch effects can obscure biological differences. Ask the provider how they handle batch structure and whether they recommend processing all samples in a single batch. If that is not possible, ask whether they will include batch information in the metadata and whether they can perform batch correction as part of the analysis.

Limitations and Interpretation Boundaries

Single-cell sequencing measures RNA abundance, not protein abundance or functional activity. A cell type identified by its transcriptome may not correspond to a functionally distinct population, and transcriptomic states can be plastic. In pancreatic cancer, for example, cell state was shown to be driven by the microenvironment, and culture models introduced strong biases in transcriptional state representation [9]. This finding underscores that single-cell data reflect the context in which cells were captured, and that ex vivo manipulation can change cell state.

Cell type annotation depends on the reference data and marker genes used. Different annotation tools and reference atlases can produce different labels for the same clusters. The provider should document which reference and which annotation method they used, and you should validate the annotation against your own knowledge of the tissue.

Single-cell data are also subject to technical dropout, where genes expressed in a cell are not detected because of limited sequencing depth. This dropout creates zeros in the count matrix that do not necessarily mean the gene is not expressed. Downstream analysis tools account for dropout in different ways, and the choice of tool affects the results.

For studies of disease, single-cell data from human tissue are often limited by sample availability and by the difficulty of obtaining appropriate controls. Studies of Parkinson's disease have shown that bulk and single-cell RNA sequencing provide complementary information, with single-cell approaches adding cell-level resolution and bulk approaches providing whole-tissue context [19]. Neither approach alone is sufficient for a complete understanding of disease mechanisms.

Safety and Regulatory Context

Single-cell sequencing of human samples involves privacy and consent considerations. If you are working with human tissue, confirm that your samples were collected with appropriate informed consent and that your study has institutional review board approval. The provider should have data security measures in place to protect identifiable information, and you should confirm that their data handling practices comply with your institution's requirements.

For studies involving genetically modified cells, such as CRISPR-edited natural killer cells, safety assessment is a separate consideration that extends beyond sequencing services [17]. Single-cell sequencing can be used to evaluate off-target effects, but the interpretation of those data requires specialized expertise. If your project involves gene editing, ask the provider whether they have experience with the specific quality control metrics needed for edited cell products.

For pathogen genomics, such as sequencing of microsporidia, the quality of the input material determines the usefulness of the sequencing data. A protocol for culturing Encephalitozoon species in human foreskin fibroblasts produced high-quality genomic DNA with minimal host contamination, with 83 to 97 percent of reads mapping to microsporidian genomes [16]. If your project involves infectious agents, confirm that the provider can handle the biosafety requirements and that their sequencing and analysis pipelines are appropriate for the organism.

Professional Escalation Criteria

You should escalate to a more experienced colleague, a core facility director, or a statistical consultant when any of the following situations arise:

• The provider reports quality control metrics that fall outside the ranges you specified, and the provider cannot explain the cause
• The provider recommends a platform or preservation method that you have not validated for your sample type
• The cost of re-sequencing or additional analysis exceeds a threshold that you defined in your project plan
• The provider cannot document the exact pipeline versions and parameters used for data processing
• The data you receive contain artifacts that the provider cannot explain, such as unexpected doublet rates or batch effects
• Your analysis plan requires tools or methods that the provider has not used before, such as SCENIC or integration with a specific public atlas
• The data sharing requirements of your funder or institution exceed what the provider can deliver

In each case, document the issue in writing and request a written response from the provider. If the provider cannot resolve the issue, consider transferring the project to a different vendor or to a core facility with more experience.

Cost Considerations and Budget Planning

Single-cell sequencing costs scale with the number of cells, the sequencing depth, and the complexity of the analysis. Per-sample costs include library preparation reagents, sequencing reagents, and labor. Sequencing costs depend on the platform and the read length, with longer reads and higher depth increasing the price.

Budget for the following items beyond the quoted per-sample price:

• Sample collection and preservation reagents
• Shipping costs, including dry ice or other temperature control
• Quality control steps that the provider may bill separately
• Re-sequencing of failed libraries
• Additional analysis beyond the basic deliverable
• Data storage fees if the provider charges for long-term retention
• Computational resources if you plan to reanalyze the data locally

Ask the provider for a total project cost estimate that includes all of these items. Compare estimates from at least three providers, and ask each provider to explain any major differences in price. A lower price may reflect lower sequencing depth, less analysis, or less rigorous quality control.

How to Evaluate Provider Validation Data

A provider that has validated its platform on your sample type will have data to show you. Ask for example quality control reports from projects similar to yours, including the cell counts, gene counts, mapping rates, and doublet rates they achieved. Ask whether they have processed your specific tissue type or cell type before, and request references from other researchers who have used their service.

Ask how the provider handles new sample types. A provider that has never processed your tissue may not know the optimal dissociation protocol or the expected cell type composition. Ask whether they will run a pilot experiment on a small number of cells before committing to the full project, and whether the pilot cost is included in the quote.

For studies that will be compared with public datasets, ask the provider whether they have experience with the specific reference atlas you plan to use. For example, if you are studying lung endothelial cells, the integrated atlas of human lung endothelial cells provides a reference for annotation [12]. The provider should be able to align your data to that reference and to explain any differences in cell type composition.

Data Reproducibility and Reporting Standards

Reproducibility in single-cell studies depends on detailed reporting of methods and quality control metrics. The FAIR Guiding Principles describe the expectations for findable, accessible, interoperable, and reusable data [4]. A provider that follows these principles will deliver data with clear metadata, standard file formats, and documentation that allows others to reuse the data.

Ask the provider whether they provide a methods report that includes the following information:

• The exact kit and chemistry version used for library preparation
• The sequencing instrument and read configuration
• The reference genome version and annotation used for alignment
• The alignment and counting software versions and parameters
• The quality control thresholds applied and the number of cells removed at each step
• The software versions used for clustering and cell type annotation

This information is essential for writing the methods section of a paper and for depositing data in public repositories. The NCBI and EMBL-EBI provide resources for data submission and for accessing training materials on data management [1][2]. If the provider cannot supply this information, the data may not be suitable for publication or for sharing.

Frequently Asked Questions

What is the difference between single-cell and single-nucleus RNA sequencing?

Single-cell RNA sequencing measures the transcriptome of intact cells, which requires a single-cell suspension. Single-nucleus RNA sequencing measures the transcriptome of isolated nuclei, which can be obtained from frozen tissue or from tissues that are difficult to dissociate. Single-nucleus approaches avoid the gene expression changes that can occur during enzymatic dissociation and are often used for brain tissue and other complex tissues. The choice depends on your sample type and your biological question.

How many cells should I sequence per sample?

The number of cells depends on the expected abundance of the cell types you want to study. If you are studying a rare cell population, you may need to sequence tens of thousands of cells to capture enough of that population. If you are studying a homogeneous population, a few thousand cells may be sufficient. The provider should help you estimate the number of cells needed based on your experimental design and the expected cell type composition.

What sequencing depth do I need?

Sequencing depth is measured in reads per cell. Higher depth increases gene detection sensitivity up to a saturation point, beyond which additional reads add little information. The optimal depth depends on the platform, the cell type, and the biological question. Ask the provider to justify their depth recommendation and to show data from their own validation runs.

How do I know if my sample quality is acceptable?

The provider should measure cell viability and cell count before library preparation. Viability thresholds vary by platform and sample type, but a common threshold is 70 to 80 percent viable cells. The provider should also assess the presence of debris, clumps, and ambient RNA. If your sample falls below the provider's threshold, they should tell you before proceeding and discuss options such as re-collection or single-nucleus sequencing.

What should I do if my data quality is poor?

First, ask the provider for the quality control report and for an explanation of any metrics that fall outside expected ranges. Common issues include high mitochondrial read fractions, low gene counts, and high doublet rates. The provider may recommend re-sequencing, re-processing, or adjusting the analysis parameters. If the provider cannot explain the issue, escalate to a more experienced colleague or consider a different provider.

Can I compare data from different providers or platforms?

Comparing data across platforms is possible but requires careful normalization and batch correction. The multisite study of preservation methods found that performance varied across platforms, which means that technical differences can be mistaken for biological differences [14]. If you plan to compare data generated by different providers, use the same platform and chemistry for all samples, or apply rigorous batch correction methods and document the limitations.

How long does a typical single-cell sequencing project take?

The timeline depends on the number of samples, the platform, and the sequencing capacity of the provider. Library preparation typically takes one to two days, sequencing takes one to several days depending on the depth and the instrument, and primary analysis takes a few days. A realistic timeline from sample receipt to data delivery is two to four weeks, but you should confirm this with the provider and ask about potential delays.

What analysis should I expect from the provider?

At a minimum, the provider should deliver raw FASTQ files, a count matrix, and a quality control report. Many providers also deliver processed objects in Seurat or Scanpy format, along with clustering and cell type annotation. Ask for a written description of the analysis scope and for an example report before you commit. Additional analysis, such as differential expression, trajectory inference, or regulatory network analysis, may be available at an additional cost.

Related Bioinformatics Guides

• Single-Cell RNA Sequencing: From Bulk to Resolution
• Master Guide: Single-Cell RNA Sequencing Bioinformatics Workflows
• Foundation Models for Single-Cell Biology
• Single-Cell ATAC-Seq Bioinformatics
• Single-Cell RNA-Seq Analysis Pipelines for Veterinary Immunology

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.
• Single-cell RNA-seq reveals ectopic and aberrant lung-resident cell populations in idiopathic pulmonary fibrosis.. Science advances, 2020.
• SCENIC: single-cell regulatory network inference and clustering.. Nature methods, 2017.
• Deciphering maternal-fetal cross-talk in the human placenta during parturition using single-cell RNA sequencing.. Science translational medicine, 2024.
• Single-cell RNA sequencing reveals placental response under environmental stress.. Nature communications, 2024.
• Microenvironment drives cell state, plasticity, and drug response in pancreatic cancer.. Cell, 2021.
• A Single-Cell Transcriptome Atlas of the Aging Drosophila Brain.. Cell, 2018.
• Single-cell analysis of human adipose tissue identifies depot and disease specific cell types.. Nature metabolism, 2020.
• Integrated Single-Cell Atlas of Endothelial Cells of the Human Lung.. Circulation, 2021.
• Best practices framework for using 16S rRNA gene sequencing in poultry microbiota research.. 2026.
• Multisite Assessment of Methods for Cell Preservation Upstream of Single-Cell RNA Sequencing.. 2026.
• The Pyrosequencing-Based Method for JAK2 Exon 12 Somatic Mutation Detection. 2026.
• In vitro culture of human-infecting Encephalitozoon spp. for genome sequencing with minimal host contaminant.. 2026.
• Engineering with care: safety assessment platforms for CRISPR-modified natural killer cells.. 2025.
• Split Pool Ligation-based Single-cell Transcriptome sequencing (SPLiT-seq) data processing pipeline comparison. BMC Genomics, 2024.
• Transcriptomics of Human Brain Tissue in Parkinson’s Disease: a Comparison of Bulk and Single-cell RNA Sequencing. Molecular Neurobiology, 2024.
• Assessment of XCI skewing and demonstration of XCI escape region based on single-cell RNA sequencing: comparison between female Grave’s disease and control. BMC Molecular and Cell Biology, 2025.
• Next-generation sequencing in the biodiversity conservation of endangered medicinal plants. Environmental Science and Pollution Research, 2022.

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