CoChat
Sign Up
Cited by 1,069Single-cell and spatial transcriptomicsGenomics and Phylogenetic StudiesInnovative Microfluidic and Catalytic Techniques Innovation

Massively parallel single-nucleus RNA-seq with DroNc-seq

Open access
Habib, Naomi, Avraham-Davidi, Inbal, Basu, Anindita, Burks, Tyler, Shekhar, Karthik, Hofree, Matan, Choudhury, Sourav R, Aguet, François +6 more·2017·Nature Methods
DOI 10.1038/nmeth.4407PMC PMC5623139

Published version of a preprint

Open this paper in CoChat — free accountOpen PDF

Abstract

Single-nucleus RNA sequencing (sNuc-seq) profiles RNA from tissues that are preserved or cannot be dissociated, but it does not provide high throughput. Here, we develop DroNc-seq: massively parallel sNuc-seq with droplet technology. We profile 39,111 nuclei from mouse and human archived brain samples to demonstrate sensitive, efficient, and unbiased classification of cell types, paving the way for systematic charting of cell atlases.

Integrity signals

Registry

Crossref · 10.1038/nmeth.4407

Cites

32 of 32 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 32 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    The authors profiled 39,111 nuclei from archived mouse and human brain samples using DroNc-seq.

    “We profile 39,111 nuclei from mouse and human archived brain samples”
    Quote in the paperSubstantial claim
  • Findingreasoning

    DroNc-seq is a droplet-based method for massively parallel single-nucleus RNA sequencing.

    “Here, we develop DroNc-seq, massively parallel sNuc-Seq with droplet technology.”
    Quote in the paperSubstantial claim
  • Findingown data

    DroNc-seq yielded high-quality libraries of neurons and glia from each human archived sample despite quality variation.

    “Although the human archived samples varied in quality, DroNc-seq yielded high-quality libraries of both neurons and glia cells from each sample”
    Quote in the paperSubstantial claim
  • Findingown data

    Empirical doublet rates ranged from about 1% for mouse tissue to about 5% for species-mixing experiments.

    “Empirical doublet rates in experiments ranged from ~1% (mouse tissue; clustering analysis) to ~5% (species-mixing).”
    Quote in the paperNot a substantial claim
  • Findingreasoning

    The smaller droplet volume in DroNc-seq gives higher mRNA concentration in drops (>5x) than 125 μm Drop-seq drops.

    “The smaller droplet volume in DroNc-seq results in higher mRNA concentration in drops (> 5x) compared to 125 μm drops in Drop-seq.”
    Quote in the paperSubstantial claim
  • May combine two claims. 23
  • Classified as probably not a substantial claim. 1

  • May combine two claims.

    Findingown data

    In 3T3 comparison, Drop-seq detected 5,134 genes per cell and DroNc-seq detected 3,295 genes per nucleus.

    “Both Drop-seq and DroNc-seq yielded high-quality libraries, detecting an average of 5,134 and 3,295 genes for cells and nuclei, respectively”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The 3T3 cell line is a suitable model for comparing single-cell and single-nucleus RNA-seq performance.

  • May combine two claims.

    Findingown data

    The nuclei isolation protocol used here is more efficient than sucrose gradient centrifugation and avoids ultra-centrifugation.

    “The nuclei isolation protocol used here is more efficient than the gradient centrifugation based method and does not require ultra-centrifugation.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Each mouse brain cluster had a distinct expression signature and was supported by nuclei from all mice.

    “Each had a distinct expression signature (Fig. 1d, Supplementary Table 4) and was supported by nuclei from all mice”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    DroNc-seq GABAergic sub-clusters matched nearly one-to-one with sNuc-Seq hippocampus sub-clusters and satisfactorily with fine-resolution visual cortex scRNA-seq sub-clusters.

    “the DroNc-seq sub-clusters matched with a nearly one-to-one mapping to sub-clusters defined by sNuc-Seq3 on the hippocampus, and with a satisfactory match to sets of fine-resolution sub-clusters defined by scRNA-seq of a distinct brain region, the visual cortex17”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [3] Habib, N. et al. Science 353, 925–928 (2016). source

      Not checked in the free run.

    • cites [17] Tasic, B. et al. Nat. Neurosci. 19, 335–346 (2016). source

      Not checked in the free run.

  • May combine two claims.

    Findingown data

    Beads at 350,000/ml and nuclei at 300,000/ml performed best, with low human-mouse doublet rate and fewer clogging events.

    “found that beads at 350,000/ml and nuclei at 300,000/ml concentrations performed best, in terms of low human-mouse doublet rate and fewer clogging events during droplet generation.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    At 160,000 reads per nucleus, DroNc-seq detected average 3,295 genes (4,643 transcripts) in 3T3 nuclei, 2,731 genes (3,653 transcripts) in mouse brain, and 1,683 genes (2,187 transcripts) in human brain.

    “DroNc-seq (for samples sequenced at 160,000 reads per nucleus, Methods) detected on average 3,295 genes (4,643 transcripts) for 3T3 nuclei, 2,731 genes (3,653 transcripts) for mouse brain, and 1,683 genes (2,187 transcripts) for human brain”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    DroNc-seq had similar throughput to Drop-seq, with high-quality nuclei efficiencies of 78% for 3T3, 89% for mouse brain, and 95% for human brain, versus 72% for Drop-seq cells.

    “DroNc-seq had similar throughput to Drop-seq with efficiencies of 78% (for 3T3 nuclei), 89% (mouse brain), and 95% (human brain)”
    • Also stated in: § Abstract · plain
      “compared to 72% high-quality cells per library”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The 3T3 cell line is a suitable model for comparing single-cell and single-nucleus RNA-seq performance.

  • May combine two claims.

    Findingown data

    DroNc-seq captured finer human cell distinctions, including CA pyramidal neuron subtypes, PFC glutamatergic subtypes with layer markers, and GABAergic subtypes.

    “These include: subtypes of CA pyramidal neurons, reflecting anatomical distinctions within the hippocampus (Supplementary Fig. 8); subtypes of glutamatergic neurons in the PFC expressing unique cortical layer marker genes, such as RORB (layer 4–54,17) (Supplementary Fig. 9, Supplementary Table 9); and subtypes of GABAergic neurons”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    DroNc-seq generated high-quality expression profiles from 5,636 3T3 nuclei, 19,561 adult frozen mouse brain nuclei, and 19,550 archived frozen adult human post-mortem brain nuclei.

    “We used DroNc-seq to robustly generate high quality expression profiles of nuclei from a mouse cell line (3T3, 5,636 nuclei), adult frozen mouse brain tissue (19,561), and archived frozen adult human post-mortem tissue (19,550 nuclei).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    GABAergic neurons of the same class from different brain regions and samples grouped together, as did non-neuronal cells.

    “GABAergic neurons of the same class but from different brain regions (and different samples) group together, as do non-neuronal cells”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Clustering 13,313 nuclei from frozen adult mouse hippocampus and PFC revealed groups corresponding to known cell types and anatomically distinct brain regions.

    “Clustering11 of 13,313 nuclei profiled from frozen adult mouse hippocampus (n=4 mice) and prefrontal cortex (PFC, n=4)”
    • Also stated in: § Abstract · plain
      “revealed groups of nuclei corresponding to known cell types (e.g., GABAergic neurons) and to anatomically distinct brain regions or sub-regions”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [11] Shekhar, K. et al. Cell 166, 1308–1323 (2016). source

      Not checked in the free run.

  • May combine two claims.

    Findingown data

    Over 84% of reads aligned to the genome in both methods; in cells 75.2% of genomic reads mapped to exons and 9.1% to introns, while in nuclei 46.2% mapped to exons and 41.8% to introns.

    “In both methods over 84% of reads align to the genome (in a representative example), but in cells 75.2% of these genomic reads map to exons and 9.1% map to introns, whereas in nuclei 46.2% of genomic reads map to exons and 41.8% to introns”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The 3T3 cell line is a suitable model for comparing single-cell and single-nucleus RNA-seq performance.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingreasoning

    DroNc-seq libraries can be re-sampled multiple times from the same input to sequence additional nuclei.

    “which can be re-sampled multiple times if a user wishes to sequence additional nuclei from the same input”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Technical replicates of 3T3 cells or nuclei had a Pearson correlation of 0.99 with a standard deviation of 0.0023, while comparisons between nuclei and cells had a correlation of 0.81 with a standard deviation of 0.0024.

    “the Pearson correlation coefficients between technical replicates of cells or nuclei (all have r=0.99+/−stdev=0.0023), and between nuclei and cells (r=0.81+/−stdev= 0.0024).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In 3T3 cell lines, differential expression testing found only two genes up-regulated in nuclei, Malat1 and Meg3, and 57 genes up-regulated in cells, including mitochondrial and ribosomal protein RNAs.

    “We found only two genes up-regulated in the nuclei (lincRNAs Malat1 and Meg3), and 57 genes up regulated in cells, including many mitochondrial RNAs and ribosomal protein RNAs (known to be stable and thus enriched in cells compared to nuclei13,14)”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The nuclei grouped into separate clouds in tSNE space that did not reflect batch; each cluster mixed cells from all technical and biological replicates and showed variable gene counts.

    “Overall, the nuclei separated into distinct point clouds in tSNE space that were not driven by batch; each cluster/cloud was an admixture of cells from all technical and biological replicates, with variable numbers of genes.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    For 3T3 nuclei, 125 μm Drop-seq and 75 μm DroNc-seq devices gave similar results, while 3T3 cells profiled by Drop-seq had better efficiency and complexity.

    “for 3T3 nuclei, both 125 μm Drop-seq and 75 μm DroNc-seq devices yielded similar results, while 3T3 cells profiled by Drop-seq had better efficiency and complexity.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The 3T3 cell line is a suitable model for comparing single-cell and single-nucleus RNA-seq performance.

  • May combine two claims.

    Interpretationown data

    DroNc-seq classifications and signatures were congruent with prior human and mouse studies despite fewer detected genes per nucleus, and were obtained with improved throughput and cost.

    “Classifications and signatures derived from DroNc-seq profiles were congruent with prior studies in human and mouse (despite the lower number of detected genes per nucleus), but were derived with significantly improved throughput and cost.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    DroNc-seq profiling of frozen archived mouse and human brain tissues identified cell types, subtypes, rare cells, expression signatures, and activated pathways.

    “DroNc-seq profiling from mouse and human frozen archived brain tissues successfully identified cell types and subtypes, rare cells, expression signatures, and activated pathways.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    DroNc-seq is robust, cost-effective, and easy to use.

    “DroNc-seq is a massively-parallel single nucleus RNA-seq method that is robust, cost-effective, and easy to use.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Massively parallel scRNA-seq methods such as Drop-seq can be applied at scale cost-effectively but require intact single-cell suspensions.

    “massively parallel scRNA-seq methods, such as Drop-seq7 and related methods8–10 can be readily applied at scale11 in a cost-effective manner12, but require intact single cell suspension as input.”
    Quote in the paperSubstantial claim0 of 6 sources checked
    • cites [7] Macosko, E.Z. et al. Cell 161, 1202–1214 (2015). source

      Not checked in the free run.

    • cites [8] Dixit, A. et al. Cell 167, 1853–1866 (2016). source

      Not checked in the free run.

    • cites [9] Adamson, B. et al. Cell 167, 1867–1882 (2016). source

      Not checked in the free run.

    • cites [10] Klein, A.M. et al. Cell 161, 1187–1201 (2015). source

      Not checked in the free run.

    • cites [11] Shekhar, K. et al. Cell 166, 1308–1323 (2016). source

      Not checked in the free run.

    • cites [12] Ziegenhain, C. et al. Cell 65, 631–643 (2017).

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    Nuclei grouped primarily by cell type rather than individual, indicating cell-type signatures are largely consistent across individuals.

    “Nuclei grouped primarily by cell type and not by individual, indicating that cell-type signatures are largely consistent across individuals.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Existing snRNA-seq methods use FACS sorting into plates or C1 microfluidics, which are not scalable to tens of thousands of nuclei or many samples.

    “However, these methods either rely on sorting nuclei by FACS into plates (96 or 384 wells)3,5 or on C1 microfluidics4, neither of which are scalable, precluding profiling tens of thousands of nuclei”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites [3] Habib, N. et al. Science 353, 925–928 (2016). source

      Not checked in the free run.

    • cites [5] Lacar, B. et al. Nat. Commun. 7, 11022 (2016). source

      Not checked in the free run.

    • cites [4] Lake, B.B. et al. Science 352, 1586–1590 (2016). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    DroNc-seq enables sensitive, efficient, and unbiased classification of cell types.

    “to demonstrate sensitive, efficient and unbiased classification of cell types”
    Quote in the paperSubstantial claim
24 of 24

  • The paper relies on this without support.

    Harsh enzymatic dissociation for brain tissue damages neurons and RNA, biases recovery of some cell types, and works only on young organisms.

    “The necessary harsh enzymatic dissociation is particularly problematic for brain tissue because it harms the integrity of neurons and their RNA, biases data in favour of recovery of some cell types, and works only on samples from young organisms”

  • The paper relies on this without support.

    The 3T3 cell line is a suitable model for comparing single-cell and single-nucleus RNA-seq performance.

    “To assess the throughput and sensitivity of DroNc-seq, we profiled 3T3 cells at the single cell (with Drop-seq) and single nucleus (with DroNc-seq) levels”

  • The paper relies on this without support.

    Neuronal and glial cells in the brain differ biologically in size and in the amount of RNA they contain.

    “Related to the number of genes, we note that there is a distinct biological difference in cell size (and expected RNA content) between neuronal and glial cells in the brain.”
3 of 3

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • snRNA-seq has unlocked the ability to unbiasedly assess transcriptional states of the human brain at single-cell resolution.

    “The advent of snRNA-seq has unlocked the ability to unbiasedly assess transcriptional states of the human brain at single-cell resolution10,11.”

    From Dissection of artifactual and confounding glial signatures by single-cell sequencing of mouse and human brain (2022) · cites this paper as [11]

  • Ribosomal protein genes are depleted in nuclei.

    “RNA of genes coding for ribosomal proteins was also depleted in nuclei (NES = −2.37, padj = 3.6e-05), as previously described (Habib et al., 2017).”

    From Single-Nucleus RNA-Seq Is Not Suitable for Detection of Microglial Activation Genes in Humans (2020) · cites this paper as Habib et al., 2017

  • snRNA-seq generates transcriptional profiles with high similarity to scRNA-seq while offering advantages: nuclei are promptly released during homogenization, minimizing cell type biases and artifactual transcriptional alteration; it is compatible with cryopreserved samples; and it can study cell types less amenable to scRNA-seq such as adipocytes and cardiomyocytes.

    “While generating transcriptional profiles with high similarity to those by scRNA-seq (Grindberg et al., 2013; Habib et al., 2017), snRNA-seq offers advantages. First, this method utilizes nuclei that are promptly released from tissues during homogenization, obviating the need for an extended digestion process and thereby minimizing cell type biases and artifactual transcriptional alteration.”

    From Robust single-nucleus RNA sequencing reveals depot-specific cell population dynamics in adipose tissue remodeling during obesity (2025) · cites this paper as Habib et al., 2017

  • The mean numbers of transcripts and genes detected per nucleus were comparable to those reported in a previous snRNA-seq study, confirming the quality of our samples prior to downstream analysis.

    “In addition, these parameters were comparable to those reported in a previous snRNA-seq study (15), confirming the quality of our samples prior to downstream analysis.”

    From Single-nucleus transcriptome analysis reveals dysregulation of angiogenic endothelial cells and neuroprotective glia in Alzheimer’s disease (2020) · cites this paper as [15]

  • Genome-wide association studies have linked mutations in Mef2c to Alzheimer's, schizophrenia, and other neurodevelopmental disorders.

    “Genome-wide association studies have linked mutations in this gene to Alzheimer’s, schizophrenia, and other neurodevelopmental disorders29.”

    From Developmental diversification of cortical inhibitory interneurons (2018) · cites this paper as [29]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

  • Reported 3T3 nucleus capture efficiency does not match its counts

    Major

    The reported 1,003 high-quality 3T3 nuclei out of 1,400 expected gives an efficiency of 71.6%, not 78%. This changes one of the stated throughput comparisons.

    Reported 78 · recomputed 71.642857

1 of 1

Cites

Flagged first, then by citations

2 of the 34 references couldn’t be matched to a known paper.

Research works printed but not linked: 2

  1. Ziegenhain, C. et al. Cell 65, 631–643 (2017).
  2. van der Maaten, L. & Hinton, G. J. Mach. Learn. Res. 9, 2579–2605 (2008).

Cited by

Sorted by citations

Related

Authors
Habib, N., Avraham-Davidi, I., Basu, A., Burks, T., Shekhar, K., Hofree, M., Choudhury, S. R., Aguet, F., Gelfand, E., Ardlie, K., Weitz, D. A., Rozenblatt-Rosen, O., Zhang, F., & Regev, A.
Affiliations
  • Broad Institute (US)
  • Harvard University (US)
Corresponding author
  • Zhang, Feng
  • Regev, Aviv
ORCID
Venue
Nature Methods
Publisher
Springer Science and Business Media LLC
Published
2017-08-28
Volume
14
Issue
10
Pages
955-958
ISSN
  • 1548-7091 (ISSN-L)
  • 1548-7105
DOI
10.1038/nmeth.4407
PMC
PMC5623139
OpenAlex
W2747841296
Semantic Scholar
789deb7c1761ade9790f4ce05e1434a5e383f0d9
Cited by
1,069
References
34
Citations by year
2026: 87 · 2025: 92 · 2024: 118 · 2023: 131 · 2022: 140 · 2021: 161 · 2020: 139 · 2019: 104 · 2018: 81 · 2017: 13 · 2015: 1
FWCI
37.56
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Methods (publishedVersion)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion)
  • PubMed (publishedVersion)
  • Europe PMC (PubMed Central) (acceptedVersion)
  • PubMed Central (acceptedVersion)
  • Digital Access to Scholarship at Harvard (DASH) (Harvard University) (submittedVersion, cc-by)
  • eScholarship (California Digital Library) (submittedVersion)
Funders
  • National Science Foundation (1310266, P30-CA14051, 1541959, 1420570)
  • Howard Hughes Medical Institute (P30-CA14051)
  • Helen Hay Whitney Foundation
  • Harvard University (DMR-1420570, 1541959)
  • Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology (P30-CA14051)
  • National Institutes of Health (HHSN268201000029C, DMR-1420570)
  • Klarman Cell Observatory, Broad Institute
  • National Heart, Lung, and Blood Institute (HHSN268201000029C, P01HL120839)
  • National Institute of Mental Health (DP1MH100706, R01MH110049, U01MH105960)
  • National Cancer Institute (CA14051, R33CA202820, P30CA014051)
  • National Institute of Allergy and Infectious Diseases (U24AI118672)
  • Division of Materials Research (DMR 14–20570, DMR-1310266, 1420570, 1310266)
  • Common Fund
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Phylogenetic Studies
  • Physical Sciences › Engineering › Biomedical Engineering › Innovative Microfluidic and Catalytic Techniques Innovation
Keywords
single-nucleus RNA sequencing, droplet-based sequencing, next-generation sequencing, postmortem brain, cell type classification, cell atlas, mouse brain, human brain
MeSH terms
Animals, Humans, RNA, Transcription, Genetic, Biomarkers, 3T3 Cells, Sequence Analysis, RNA, Principal Component Analysis, Mice, HEK293 Cells, Single-Cell Analysis

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-07.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-07.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review