Single Nuclear RNA Sequencing: Methods and Applications

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-11

Key Takeaways

• Single nuclear RNA sequencing (snRNA-seq) profiles the transcriptome of individual nuclei, offering compatibility with frozen and post-mortem tissues, thereby enabling analysis of archival clinical samples and post-mortem brain tissue where intact cell viability is compromised.
• The method circumvents enzymatic dissociation, a common step in single-cell RNA sequencing (scRNA-seq), thereby mitigating stress-induced transcriptional artifacts such as the upregulation of immediate early genes (FOS, JUN) and heat shock proteins.
• Nuclei isolation involves hypotonic lysis and density gradient centrifugation, followed by droplet-based microfluidics for barcoded library preparation, capturing nascent pre-mRNAs and nuclear-retained transcripts, though with a lower overall gene detection rate compared to scRNA-seq.
• snRNA-seq is crucial for profiling cell types notoriously difficult to dissociate into single cells, including neurons with their complex morphology, large adipocytes, and multinucleated cardiomyocytes, by eliminating the requirement for intact cell membranes.
• Key applications include the construction of comprehensive brain cell atlases, identification of disease-associated cell states in neurodegenerative disorders (e.g., reactive astrocytes in Alzheimer's disease), and profiling of tissues like heart and kidney.
• Challenges in snRNA-seq data analysis include ambient RNA contamination, doublet detection, batch effects, and the need for specialized computational tools to accurately quantify intronic reads and filter low-quality nuclei based on metrics like UMI counts and mitochondrial RNA fraction.

Introduction to Single Nuclear RNA Sequencing

What is snRNA-seq?

Single nuclear RNA sequencing (snRNA-seq) is a high-throughput transcriptomic technique that profiles the RNA content of individual nuclei isolated from tissues. Unlike conventional bulk RNA-seq, which measures average gene expression across millions of cells, snRNA-seq captures the transcriptional state of single nuclei, enabling researchers to resolve cellular heterogeneity within complex tissues at single-nucleus resolution.

The method relies on the fact that the nucleus contains a substantial fraction of the cellular transcriptome, specifically, nascent (unspliced) pre-mRNA transcripts, mature mRNA molecules that have not yet been exported to the cytoplasm, and various non-coding RNAs. By isolating intact nuclei and sequencing their RNA content, snRNA-seq provides a snapshot of transcriptional activity that is largely comparable to that obtained from whole-cell approaches.

The fundamental distinction between snRNA-seq and single-cell RNA sequencing (scRNA-seq) lies in the input material. scRNA-seq requires intact, viable whole cells, typically obtained through enzymatic dissociation of fresh tissue. snRNA-seq, in contrast, uses isolated nuclei, which can be extracted from fresh or flash-frozen tissue samples. This seemingly minor difference has profound implications for experimental design, sample compatibility, and the types of biological questions that can be addressed.

Why use nuclei instead of cells?

The choice of nuclei over whole cells is driven by several practical and biological considerations. First, nuclei are considerably more robust than whole cells. They withstand mechanical shearing, cryopreservation, and prolonged storage better than intact cells, which are prone to lysis and transcriptional perturbation during isolation. This robustness makes snRNA-seq the method of choice for archived clinical specimens, post-mortem brain tissue, and other samples that cannot be processed immediately.

Second, enzymatic dissociation, the standard method for obtaining single-cell suspensions, activates stress-responsive genes, including immediate early genes (e.g., FOS, JUN, EGR1) and heat shock proteins (e.g., HSPA1A, HSP90AB1). This dissociation-induced transcriptional artifact can confound downstream analyses, particularly when studying neuronal activity or stress responses. Nuclei isolation, which relies on hypotonic lysis and density gradient centrifugation rather than enzymatic digestion, minimizes this artifact because the nuclear transcriptome is less affected by cytoplasmic signaling cascades triggered by receptor engagement.

Third, certain cell types are notoriously difficult to dissociate into viable single cells. Neurons, with their elaborate dendritic arbors and long axonal projections, are particularly susceptible to damage during mechanical trituration. Adipocytes, which are large and lipid-laden, float during centrifugation and are lost from standard cell suspensions. Cardiomyocytes are multinucleated and too large for most microfluidic platforms. snRNA-seq circumvents these issues by eliminating the need for intact cell membranes.

Finally, snRNA-seq captures a greater proportion of the transcriptome in certain contexts. While cytoplasmic mRNA is lost during nuclei isolation, the nuclear fraction is enriched for nascent transcripts and long non-coding RNAs, including Small Nuclear RNA species involved in splicing. This enrichment can be advantageous for studying transcriptional regulation and splicing dynamics.

The snRNA-seq Workflow

Nuclei isolation and purification

The snRNA-seq workflow begins with tissue disruption and nuclei isolation. The specific protocol depends on the tissue type, but the general principles are consistent across applications.

For fresh or flash-frozen tissue, the sample is first minced into small pieces (1–2 mm³) using a scalpel or razor blade on a chilled surface. The minced tissue is then homogenized in a hypotonic lysis buffer containing a mild detergent to disrupt cell membranes while preserving nuclear integrity. A typical lysis buffer contains 10 mM Tris-HCl (pH 7.4), 10 mM NaCl, 3 mM MgCl₂, 0.1% Nonidet P-40 (NP-40) or IGEPAL CA-630, and a ribonuclease (RNase) inhibitor such as SUPERase•In at 1 U/µL. The low salt concentration causes cells to swell and burst, while the detergent solubilizes the plasma and organellar membranes. Importantly, the buffer lacks calcium chelators like EDTA, because magnesium is required to maintain nuclear structure.

The homogenate is then layered onto a sucrose cushion (typically 1.8 M sucrose in lysis buffer) and centrifuged at high speed (approximately 24,000 × g for 2 hours at 4°C) in a swinging-bucket rotor. Nuclei, being denser than cellular debris, pellet through the sucrose layer, while cytoplasmic components and membrane fragments remain at the interface. Alternatively, a discontinuous iodixanol gradient (OptiPrep) can be used, which requires shorter centrifugation times (20 minutes at 10,000 × g) and is gentler on fragile nuclei.

After centrifugation, the nuclear pellet is resuspended in a wash buffer containing 1× phosphate-buffered saline (PBS), 0.01% bovine serum albumin (BSA), and an RNase inhibitor. The nuclei are then counted using a hemocytometer or automated cell counter, and their integrity is assessed by staining with a nuclear dye such as 4′,6-diamidino-2-phenylindole (DAPI) or SYTOX Green. Intact nuclei appear as round, uniformly stained particles with minimal debris. Flow cytometry or fluorescence-activated nuclei sorting (FANS) can be used to further purify nuclei based on DNA content or specific nuclear markers, which is particularly useful for isolating neuronal nuclei using antibodies against NeuN (RBFOX3).

For droplet-based platforms such as the 10x Genomics Chromium system, the nuclear suspension is adjusted to a concentration of approximately 700–1,200 nuclei/µL to achieve a target capture rate of 5,000–10,000 nuclei per reaction.

Library preparation and sequencing

The purified nuclei are then processed through a droplet-based microfluidic platform. In the 10x Genomics Chromium system, individual nuclei are co-encapsulated with barcoded gel beads in nanoliter-scale droplets (GEMs, or Gel Bead-in-EMulsion). Each gel bead carries a unique 16-nucleotide (nt) cell barcode, a 10-nt unique molecular identifier (UMI), and a poly(dT) primer with an Illumina sequencing adapter.

Within each droplet, the nuclear membrane is permeabilized, and reverse transcription is initiated. The poly(dT) primer anneals to the poly(A) tail of mature mRNAs, and Moloney murine leukemia virus (MMLV) reverse transcriptase synthesizes complementary DNA (cDNA). The reaction is carried out at 42°C for 2 hours in the presence of a template-switching oligonucleotide (TSO), which adds a common sequence to the 5′ end of the cDNA. This template-switching mechanism enables full-length cDNA synthesis and subsequent amplification.

Following reverse transcription, the droplets are broken, and the cDNA is pooled and amplified by PCR. The amplification step typically uses 12–14 cycles of PCR with primers that add the Illumina P5 and P7 sequences, as well as sample indices. The resulting library is then purified using SPRI beads (e.g., AMPure XP) to remove primers and short fragments, and the quality is assessed using a Bioanalyzer or TapeStation. A typical snRNA-seq library shows a broad peak between 300 and 600 bp, corresponding to cDNA fragments derived from the 3′ ends of transcripts.

Sequencing is performed on an Illumina platform (NovaSeq 6000, NextSeq 2000, or similar). The recommended sequencing depth is approximately 20,000–50,000 read pairs per nucleus, with a read structure of 28 bp for Read 1 (containing the cell barcode and UMI), 8 bp for the i7 index, and 90 bp for Read 2 (containing the cDNA insert). Deeper sequencing yields diminishing returns because the UMI-based counting saturates; most genes are detected with 20,000 reads per nucleus, and additional reads primarily increase the count depth for highly expressed genes.

Data Analysis Pipeline for snRNA-seq

Quality control metrics

The raw sequencing data are processed using a pipeline that begins with demultiplexing and alignment. The 10x Genomics Cell Ranger software is the most commonly used tool for initial processing. It performs the following steps:

1. Demultiplexing: Reads are assigned to individual nuclei based on their cell barcode sequences.
2. UMI deduplication: Reads with the same UMI and gene annotation are collapsed into a single transcript count, eliminating PCR amplification bias.
3. Alignment: Reads are aligned to the reference genome using a splice-aware aligner (STAR). For snRNA-seq, the reference includes both exonic and intronic sequences, because nuclear RNA contains substantial amounts of unspliced pre-mRNA.
4. Gene counting: A count matrix is generated, with rows representing genes and columns representing nuclei.

Quality control (QC) metrics are then applied to filter low-quality nuclei. The key metrics include:

• Total UMI counts: Nuclei with very low UMI counts (<500) likely represent empty droplets or damaged nuclei with degraded RNA. Nuclei with extremely high UMI counts (>50,000) may be doublets (two nuclei captured in one droplet).
• Number of detected genes: Similar to UMI counts, this metric distinguishes real nuclei from background. A typical threshold is 200–500 genes per nucleus.
• Mitochondrial RNA fraction: In scRNA-seq, a high fraction of mitochondrial reads (>20%) indicates dying or lysed cells. In snRNA-seq, this metric is less informative because mitochondria are excluded from the nuclear preparation. However, a very low mitochondrial fraction (<0.5%) is expected, and elevated levels may indicate cytoplasmic contamination.
• Nuclear marker genes: Expression of nuclear-enriched genes such as MALAT1 (a long non-coding RNA retained in the nucleus) and NEAT1 can confirm the nuclear origin of the captured RNA.

After filtering, the data are normalized. The most common approach is log-normalization, where each nucleus's UMI counts are divided by the total UMI count, multiplied by a scale factor (typically 10,000), and log-transformed. This accounts for differences in sequencing depth between nuclei.

Clustering and cell type identification

Following normalization, the data are subjected to dimensionality reduction. Principal component analysis (PCA) is first applied to the most variable genes (typically 2,000–5,000 genes selected by dispersion). The top 20–50 principal components are then used as input for graph-based clustering, implemented in algorithms such as Louvain or Leiden clustering. These algorithms construct a k-nearest-neighbor graph and partition nuclei into clusters based on transcriptomic similarity.

The resulting clusters are annotated by examining the expression of known cell-type-specific marker genes. For example, in brain tissue, SLC17A7 (VGLUT1) marks excitatory neurons, GAD1 and GAD2 mark inhibitory neurons, AQP4 marks astrocytes, MOG marks oligodendrocytes, CX3CR1 marks microglia, and PECAM1 marks endothelial cells. This annotation step requires biological knowledge and is often guided by reference atlases.

Batch correction is a critical step when integrating data from multiple samples, donors, or experimental batches. Methods such as Harmony, Seurat's integration workflow, or Scanorama align nuclei across batches by identifying shared cell states and removing technical variation. This is particularly important for snRNA-seq studies that combine data from multiple post-mortem brain samples, where post-mortem interval and tissue processing conditions introduce batch effects.

Differential expression analysis between clusters or conditions is performed using statistical tests appropriate for single-cell data, such as the Wilcoxon rank-sum test, MAST, or DESeq2 applied to pseudobulk aggregates. Pseudobulk approaches, which sum counts across all nuclei in a sample before differential testing, are preferred for comparing conditions across biological replicates because they account for inter-individual variability.

Advantages and Limitations of snRNA-seq

When to choose snRNA-seq over scRNA-seq

The choice between snRNA-seq and scRNA-seq depends on the biological question and the nature of the sample. snRNA-seq is the preferred method in the following scenarios:

snRNA-seq is particularly advantageous for studying the brain, where post-mortem tissue is often the only available source. The ability to profile archived frozen samples has enabled large-scale studies of neurological and psychiatric disorders using biobank specimens. Additionally, snRNA-seq is well-suited for tissues that are difficult to dissociate, such as heart, kidney, and adipose tissue.

Challenges in snRNA-seq data

Despite its advantages, snRNA-seq has several limitations that must be considered. The most significant is the lower gene detection rate compared to scRNA-seq. Because nuclear RNA represents only a fraction of the total cellular transcriptome, snRNA-seq typically detects 2,000–5,000 genes per nucleus, compared to 3,000–8,000 genes per cell in scRNA-seq. This reduced sensitivity can hinder the detection of low-abundance transcripts, particularly those encoding cytokines, chemokines, and immediate early genes.

The loss of cytoplasmic RNA also means that snRNA-seq cannot capture the translational state of the cell. mRNA localization to the cytoplasm, local translation at synapses, and RNA modifications that affect stability are invisible to snRNA-seq. For studies focused on post-transcriptional regulation, scRNA-seq or other approaches may be more appropriate.

Another challenge is the presence of ambient RNA, transcripts released from damaged nuclei during isolation that contaminate the droplet suspension. This ambient RNA can create spurious signals, particularly for highly expressed genes, and can obscure rare cell types. Computational methods such as SoupX or CellBender can estimate and remove ambient RNA contamination, but they require careful parameter tuning.

Finally, snRNA-seq data contain a higher proportion of intronic reads compared to scRNA-seq, because nascent pre-mRNA is enriched in the nucleus. While this can be advantageous for studying splicing and transcriptional activity, it complicates gene quantification. Most pipelines include intronic reads in gene counts, but this choice must be consistent across comparisons.

Applications in Neuroscience and Beyond

Brain cell atlas projects

The most prominent application of snRNA-seq has been in neuroscience, where it has enabled the construction of comprehensive cell atlases of the mammalian brain. The ability to profile post-mortem human brain tissue has been transformative, as fresh human brain tissue is rarely available for research.

The Allen Institute for Brain Science and the BRAIN Initiative Cell Census Network (BICCN) have used snRNA-seq to generate transcriptomic maps of the mouse and human cortex. These studies have revealed an unexpected diversity of neuronal subtypes. For example, snRNA-seq of the human primary motor cortex identified 75 transcriptionally distinct cell types, including 46 neuronal subtypes. Excitatory neurons in layer 5 were found to be particularly heterogeneous, with subtypes distinguished by expression of genes such as RORB, THEMIS, CARM1P1, and FREM3.

The human brain atlas efforts have also revealed species-specific differences in cell types. Comparison of human and mouse cortical snRNA-seq data showed that while the major cell classes (excitatory neurons, inhibitory neurons, astrocytes, oligodendrocytes, microglia) are conserved, the proportions and gene expression profiles of subtypes differ substantially. For instance, human layer 4 excitatory neurons express higher levels of RORB and lower levels of CUX2 compared to their mouse counterparts.

Disease-related studies

snRNA-seq has been extensively applied to study neurodegenerative diseases, where post-mortem brain tissue is the primary sample source. In Alzheimer's disease (AD), snRNA-seq studies have identified disease-associated cell states, including a reactive astrocyte subtype characterized by upregulation of GFAP, VIM, and SERPINA3, and a microglial state marked by expression of TREM2, APOE, and CST7. These disease-associated populations are enriched in AD brains and correlate with pathological hallmarks such as amyloid-beta plaques and neurofibrillary tangles.

In Parkinson's disease, snRNA-seq of the substantia nigra has revealed selective vulnerability of specific dopaminergic neuron subtypes. A population of dopaminergic neurons expressing SOX6 and ALDH1A1 was found to be preferentially lost in Parkinson's disease, while a neighboring population expressing CALB1 was relatively spared. This finding has implications for understanding the selective vulnerability that has puzzled neurologists for decades.

Beyond the brain, snRNA-seq has been applied to study the heart, kidney, liver, and lung. In the heart, snRNA-seq has identified distinct populations of cardiomyocytes, fibroblasts, endothelial cells, and immune cells, and has revealed how these populations change during heart failure. In the kidney, snRNA-seq has mapped the cellular composition of the nephron and has identified novel populations of collecting duct cells and podocytes. These studies demonstrate the broad utility of snRNA-seq across organ systems.

Key Evidence and Landmark Studies

Early proof-of-concept studies

The first demonstration that snRNA-seq could reliably capture the transcriptome of individual nuclei came from studies using the Drop-seq and 10x Genomics platforms. In 2017, a landmark study applied snRNA-seq to post-mortem human brain tissue and showed that nuclear transcriptomes could distinguish major cell types (neurons, astrocytes, oligodendrocytes, and microglia) with accuracy comparable to scRNA-seq. The study also demonstrated that snRNA-seq data from frozen tissue were highly reproducible across technical replicates, with correlation coefficients above 0.9.

A critical validation came from studies comparing snRNA-seq and scRNA-seq on the same tissue. In the mouse cortex, parallel snRNA-seq and scRNA-seq experiments identified the same major cell types and similar proportions of each type. However, snRNA-seq detected fewer genes per nucleus and showed reduced expression of activity-dependent genes such as FOS and ARC, confirming that nuclear RNA captures a quieter, less perturbed transcriptional state.

Large-scale atlas efforts

The Human Cell Atlas (HCA) project has incorporated snRNA-seq as a core technology for profiling tissues that are difficult to dissociate. The brain, in particular, has been a focus of large-scale snRNA-seq efforts. The BICCN generated a comprehensive atlas of the mouse primary motor cortex using a combination of snRNA-seq, scRNA-seq, and spatial transcriptomics, integrating data from over 500,000 cells to define 116 transcriptomic cell types.

In 2023, a landmark study profiled over 3 million nuclei from the human brain across multiple cortical regions and developmental stages. This study identified over 100 distinct cell types and mapped their spatial distribution using complementary approaches. The integration of snRNA-seq with genome-wide association study (GWAS) data revealed that genetic variants associated with schizophrenia, bipolar disorder, and major depression are enriched in specific neuronal subtypes, providing a cellular context for psychiatric genetics.

The Genotype-Tissue Expression (GTEx) project has also incorporated snRNA-seq to profile cell-type-specific expression across multiple human tissues. This resource has enabled researchers to link genetic variants to cell-type-specific gene expression changes, advancing our understanding of the molecular basis of complex traits.

Common Pitfalls and Troubleshooting

Ambient RNA and contamination

Ambient RNA is one of the most pervasive artifacts in snRNA-seq. During nuclei isolation, some nuclei inevitably lyse, releasing their RNA into the suspension. This RNA is then captured in droplets along with intact nuclei, contributing counts that do not reflect the true transcriptional state of the captured nucleus.

The severity of ambient RNA contamination varies with tissue type and isolation protocol. Tissues with high RNA content, such as liver and pancreas, are particularly prone to this artifact. Highly expressed genes (e.g., ALB in liver, INS in pancreas) can appear in all nuclei regardless of cell type, creating false signals.

Detection of ambient RNA contamination can be performed by examining the expression of known marker genes in cell types where they should not be expressed. For example, if ALB (a hepatocyte marker) is detected in endothelial nuclei, ambient contamination is likely. Computational tools such as SoupX estimate the ambient RNA profile from empty droplets (droplets without nuclei) and subtract its contribution from the count matrix. CellBender, a deep learning-based method, performs a similar function with improved accuracy.

To minimize ambient RNA experimentally, several strategies are effective: (1) include a high concentration of BSA (0.1%) in wash buffers to block non-specific binding, (2) perform multiple washes of the nuclear pellet, (3) use fluorescence-activated nuclei sorting (FANS) to remove debris and damaged nuclei, and (4) process samples quickly to minimize the time nuclei spend in suspension.

Doublet detection and removal

Doublets, droplets containing two or more nuclei, are a major source of spurious cell types in snRNA-seq. Doublets can form when nuclei aggregate during loading or when two nuclei are co-encapsulated by chance. The doublet rate scales with loading density; at a loading density of 10,000 nuclei per reaction, the expected doublet rate is approximately 4–5%.

Doublets are problematic because they generate artificial transcriptomes that combine the expression profiles of two distinct cell types. In clustering analyses, doublets often appear as intermediate states between two clusters, creating false "transitional" populations. They can also obscure rare cell types by diluting their signal.

Computational doublet detection methods fall into two categories. The first category, including DoubletFinder and scDblFinder, simulates artificial doublets by combining the transcriptomes of random nucleus pairs and trains a classifier to distinguish real doublets from singletons. The second category, including demuxlet and Vireo, uses genetic variation (natural or introduced) to identify doublets. For demuxlet, nuclei from multiple donors are pooled in a single reaction, and the genotype of each nucleus is inferred from the RNA reads. Nuclei with mixed genotypes are identified as doublets.

Experimental strategies to reduce doublets include: (1) optimizing the nuclei concentration to target a lower capture rate, (2) filtering nuclei suspensions through a 40 µm cell strainer to remove aggregates, and (3) using FANS to sort single nuclei into wells. The trade-off is that lower loading densities reduce throughput, so a balance must be struck based on the experimental goals.

Batch effects

Batch effects, systematic technical variation between samples processed at different times or in different conditions, are a persistent challenge in snRNA-seq studies. Sources of batch effects include differences in tissue quality, nuclei isolation efficiency, library preparation, and sequencing depth.

Batch effects can be detected by examining whether nuclei from the same biological sample cluster together in a principal component analysis before batch correction. If samples form distinct clusters that do not correspond to known biological differences, batch effects are present.

Several computational methods can correct for batch effects while preserving biological variation. Harmony iteratively clusters nuclei and learns a correction vector for each batch. Seurat's integration workflow uses canonical correlation analysis to identify shared cell states across batches and anchors the datasets together. Scanorama uses a panoramic stitching approach to merge datasets. Each method has strengths and weaknesses, and the choice depends on the data structure and the number of batches.

A critical caveat is that batch correction can over-correct, removing genuine biological differences. This is particularly problematic when the biological condition of interest (e.g., disease versus control) is confounded with batch. To mitigate this, experimental design should balance conditions across batches, and validation of corrected data should be performed by checking that known marker genes remain differentially expressed.

Summary and Future Directions

Emerging trends

Several emerging technologies are poised to extend the capabilities of snRNA-seq. Multi-omics approaches now enable simultaneous profiling of RNA and other modalities from the same nucleus. For example, snATAC-seq (single-nucleus assay for transposase-accessible chromatin) profiles chromatin accessibility alongside RNA expression, providing a joint view of gene regulation. Methods such as 10x Multiome combine snRNA-seq and snATAC-seq in a single workflow, enabling the construction of gene regulatory networks that link transcription factors to their target genes.

Spatial snRNA-seq is another frontier. While standard snRNA-seq loses spatial information, spatial transcriptomics methods such as Slide-seq and Visium can be adapted to capture nuclear RNA with positional information. These approaches are beginning to reveal how cell types are organized in tissue and how cell-cell interactions shape gene expression.

Finally, the application of snRNA-seq to clinical samples is expanding. The ability to profile archived formalin-fixed, paraffin-embedded (FFPE) tissue, albeit with reduced sensitivity, is opening new avenues for retrospective studies of disease. As the technology continues to improve, snRNA-seq is likely to become a standard tool in both basic research and clinical diagnostics.

Frequently Asked Questions

What is the difference between single nuclear RNA-seq and single-cell RNA-seq?

Single nuclear RNA-seq (snRNA-seq) profiles the RNA content of individual nuclei, while single-cell RNA-seq (scRNA-seq) profiles the RNA content of intact whole cells. snRNA-seq captures nuclear RNA, including nascent pre-mRNA and nuclear-retained transcripts, whereas scRNA-seq captures both nuclear and cytoplasmic RNA. The two methods differ in their sample requirements: snRNA-seq works with frozen or post-mortem tissue, while scRNA-seq requires fresh, viable cells.

Why use nuclei instead of whole cells for RNA sequencing?

Nuclei are used because they are more robust than whole cells, surviving freezing, thawing, and mechanical disruption that would destroy intact cells. Nuclei isolation avoids enzymatic digestion, which can activate stress-response genes and introduce transcriptional artifacts. Additionally, some cell types (such as neurons, adipocytes, and cardiomyocytes) are difficult or impossible to dissociate into viable single cells, making nuclei the only practical option.

Can snRNA-seq detect all genes expressed in a cell?

No. snRNA-seq detects only a fraction of the genes expressed in a cell, typically 2,000–5,000 genes per nucleus compared to 3,000–8,000 genes per cell in scRNA-seq. Cytoplasmic mRNAs are lost during nuclei isolation, and low-abundance transcripts may fall below the detection limit. However, snRNA-seq is enriched for nuclear-retained transcripts, including long non-coding RNAs and nascent pre-mRNAs, which may be underrepresented in scRNA-seq.

What types of samples are suitable for snRNA-seq?

snRNA-seq is suitable for fresh tissue, flash-frozen tissue, and post-mortem tissue. It is particularly valuable for archived samples that cannot be processed immediately. The method works across a wide range of tissues, including brain, heart, kidney, liver, lung, and adipose tissue. Samples should be stored at −80°C and processed with minimal freeze-thaw cycles to preserve RNA integrity.

How do you isolate nuclei for snRNA-seq?

Nuclei are isolated by homogenizing tissue in a hypotonic lysis buffer containing a mild detergent (e.g., NP-40), followed by density gradient centrifugation through sucrose or iodixanol. The nuclear pellet is washed, counted, and assessed for integrity using DAPI staining. For some applications, fluorescence-activated nuclei sorting (FANS) is used to purify specific nuclear populations.

What are the main challenges in analyzing snRNA-seq data?

The main challenges include ambient RNA contamination, doublet detection, batch effects, and the lower gene detection rate compared to scRNA-seq. Ambient RNA can create spurious signals, doublets can generate artificial cell types, and batch effects can obscure biological differences. These challenges require careful quality control, computational correction, and robust experimental design.

Is snRNA-seq more expensive than scRNA-seq?

The cost of snRNA-seq is comparable to scRNA-seq on a per-sample basis. The library preparation and sequencing costs are similar, and the main difference lies in the nuclei isolation step, which adds a small amount of time and reagent cost. However, snRNA-seq can be more cost-effective overall because it enables the use of frozen samples, which are often more readily available than fresh tissue.

Key Takeaways

• snRNA-seq profiles the transcriptome of individual nuclei and is compatible with frozen and post-mortem tissue, making it ideal for clinical and archival samples.
• The method avoids enzymatic dissociation, reducing stress-induced transcriptional artifacts that confound scRNA-seq.
• Nuclei are isolated by hypotonic lysis and density gradient centrifugation, then encapsulated in droplets for barcoded library preparation.
• snRNA-seq detects fewer genes per nucleus than scRNA-seq but captures nuclear-enriched transcripts, including nascent pre-mRNA and long non-coding RNAs.
• Major applications include brain cell atlas construction, neurodegenerative disease research, and profiling of difficult-to-dissociate tissues like heart and kidney.
• Key challenges include ambient RNA contamination, doublets, batch effects, and reduced sensitivity for low-abundance transcripts.
• Emerging multi-omics and spatial approaches are extending snRNA-seq to simultaneously profile chromatin accessibility and spatial organization.

Further Reading

• Purice MD et al. Molecular profiling of adult C. elegans glia across sexes by single-nuclear RNA-seq. Developmental cell. 2025. PubMed 40527319
• Xu Y et al. Integrated single-nuclear RNA sequencing analysis reveals distinct characteristics of mucinous adenocarcinoma in right-sided colon cancer. International journal of biological macromolecules. 2025. PubMed 40180102
• Kim HJ et al. Nuclear oligo hashing improves differential analysis of single-cell RNA-seq. Nature communications. 2022. PubMed 35562344

Related Topics

• RNA Degradation
• Piwi RNA
• RNA Localization
• RNA Location

Related Clinical & Scientific Guides

• MAPK Pathway: Mechanism, Function, and Clinical Relevance
• Mammalian Cell Culture Bioreactors: A Practical Guide
• Nucleotide Formation: Biosynthesis and Assembly of DNA/RNA Building Blocks