CoChat
Sign Up
Cited by 245Single-cell and spatial transcriptomicsGenomics and Chromatin DynamicsCRISPR and Genetic Engineering

Targeted Perturb-seq enables genome-scale genetic screens in single cells

Open access
Schraivogel, Daniel, Gschwind, Andreas R., Milbank, Jennifer H., Leonce, Daniel R., Jakob, Petra, Mathur, Lukas, Korbel, Jan O., Merten, Christoph A. +2 more·2020·Nature Methods
DOI 10.1038/s41592-020-0837-5PMC PMC7610614
Open this paper in CoChat — free account

Abstract

The transcriptome contains rich information on molecular, cellular and organismal phenotypes. However, experimental and statistical limitations constrain sensitivity and throughput of genetic screening with single-cell transcriptomics readout. To overcome these limitations, we introduce targeted Perturb-seq (TAP-seq), a sensitive, inexpensive and platform-independent method focusing single-cell RNA-seq coverage on genes of interest, thereby increasing the sensitivity and scale of genetic screens by orders of magnitude. TAP-seq permits routine analysis of thousands of CRISPR-mediated perturbations within a single experiment, detects weak effects and lowly expressed genes, and decreases sequencing requirements by up to 50-fold. We apply TAP-seq to generate perturbation-based enhancer-target gene maps for 1,778 enhancers within 2.5% of the human genome. We thereby show that enhancer-target association is jointly determined by three-dimensional contact frequency and epigenetic states, allowing accurate prediction of enhancer targets throughout the genome. In addition, we demonstrate that TAP-seq can identify cell subtypes with only 100 sequencing reads per cell.

Integrity signals

Registry

Crossref · 10.1038/s41592-020-0837-5

Cites

58 of 58 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 38 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    TAP-seq was used to build perturbation-based maps linking enhancers to target genes for 1,778 enhancers covering 2.5% of the human genome.

    “We apply TAP-seq to generate perturbation-based enhancer-target gene maps for 1,778 enhancers within 2.5% of the human genome.”
    Quote in the paperSubstantial claim
  • Findingown data

    TAP-seq can distinguish cell subtypes using only 100 sequencing reads per cell.

    “we demonstrate that TAP-seq can identify cell subtypes with only 100 sequencing reads per cell.”
    Quote in the paperSubstantial claim
  • Findingnone given

    TAP-seq reduces sequencing needs by as much as 50-fold.

    “decreases sequencing requirements up to 50-fold”
    Quote in the paperSubstantial claim
  • Findingown data

    Hits were split into 36 strong enhancer-target pairs supported by at least half of the targeting gRNAs and 45 weak pairs.

    “we further classified our hits into 36 ‘strong’ ETPs supported by at least 50% of the gRNAs targeting the enhancer, and 45 ‘weak’ ETPs”
    Quote in the paperNot a substantial claim
  • Findingown data

    Off-target amplicons and byproducts were uncommon.

    “Thus, byproducts and off-target amplicons were rare”
    Quote in the paperSubstantial claim
  • May combine two claims. 36
  • Classified as probably not a substantial claim. 12

  • May combine two claims.

    Findingreasoning

    TAP-seq detects weak effects and genes with low expression.

    “detects weak effects and lowly expressed genes”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    A model trained on chromosome 11 predicted enhancer-target pairs on chromosome 8 well, and the reverse also worked.

    “A model trained on data from chromosome 11 efficiently predicted ETPs on chromosome 8, and vice versa”
    Quote in the paperNot a substantial claim

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

    Findingown data

    For panel d, primer design succeeded for 184 out of 192 target genes.

    “Primers were successfully designed for 184 of the 192 target genes.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    For CCNE2 and PHF20L1, TAP-seq had lower statistical power than Perturb-seq; these genes had disproportionately low coverage in TAP-seq, suggesting lower amplification efficiency.

    “For two genes (CCNE2 and PHF20L1), statistical power in TAP-seq was lower than in Perturb-Seq; these genes had disproportionally little coverage in TAP-seq, indicative of a lower amplification efficiency”
    Quote in the paperNot a substantial claim

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

    Findingown data

    For two genes, TAP-seq detected perturbation effects less efficiently.

    “Two genes for which perturbation effects were detected with a lower efficiency in TAP-seq are highlighted in red.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The study found 81 significant cis-acting enhancer-target pairs, covering 24-32% of profiled genes in each region and 4.4% of tested enhancers.

    “We observed a total of 81 significant cis-acting ETPs, involving 24-32% of the genes profiled within the respective genomic region, and 4.4% of the tested enhancers”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Enhancers mainly regulate genes located near them.

  • May combine two claims.

    Findingown data

    In the enhancer screen, 93% of gRNA consensus sequences on chromosome 8 and 95% on chromosome 11 had no mismatches with the template sequence.

    “Levenshtein edit distance between the consensus sequence of a gRNA in a given cell, and the template sequence, showing that in 93 % (chr. 8) or 95 % (chr. 11) of cases, there were no mismatches between consensus and template.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    TAP-seq yields libraries ready for Illumina sequencing that match 10X Genomics single-cell 3' libraries and rely on standard Illumina Read 1 and Read 2 primers.

    “TAP-seq produces Illumina sequencing ready libraries identical to 10X Genomics single-cell 3’ libraries using standard Illumina Read 1 and Read 2 primers.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In TAP-seq libraries, prominent bioanalyzer peaks matched genes that are highly expressed in the primer panel (HBG1, HBG2, HBE1), and this was confirmed by sub-cloning and Sanger sequencing.

    “Strong peaks in TAP-seq profile correspond to highly expressed genes in the primer panel (HBG1, HBG2, HBE1), as validated by sub-cloning of bands and Sanger sequencing (not shown).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    In cross-validation, panels a, b, and d classified bone marrow cell types with similar performance (Cohen's Kappa 0.8-0.85), whereas panel c performed worse (Cohen's Kappa 0.65).

    “While a, b and d provided a similar well type classification performance (Cohen’s Kappa 0.8-0.85 in 10-fold CV), c performed worse (Cohen’s Kappa 0.65).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    On average, 37 cells per gRNA or 143 cells per enhancer were profiled, totaling 7,055 gRNA perturbations in 231,667 cells.

    “An average of 37 cells per gRNA or 143 cells per enhancer were profiled, for a total of 7,055 gRNA perturbations in 231,667 cells”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Whole transcriptome Perturb-seq covered each gRNA with a median of 415.5 cells and 8 million reads.

    “covering each gRNA with a median of 415.5 cells and 8 million reads (Figure 2b)”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    A model trained on all study data predicted enhancer-target pairs from a Perturb-seq study at 75% recall and 40% precision, but the reverse prediction was ineffective.

    “When trained on all data from this study, our model predicted ETPs identified in a Perturb-seq study9 (75% recall at a precision of 40%); the converse did not work effectively”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [9] Gasperini, M. et al. A genome-wide framework for mapping gene regulation via cellular genetic screens. Cell 176, 377–390 (2019). source

      Not checked in the free run.

  • May combine two claims.

    Findingown data

    In the TAP-seq dataset, each gRNA was covered by a median of 268.5 cells and 0.8 million reads.

    “each gRNA was covered by a median of 268.5 cells and 0.8 million reads (Figure 2b)”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    TAP-seq recovered gRNA identity at 95% efficiency, versus 39% for Perturb-seq and 89% for Perturb-seq with targeted gRNA amplification.

    “TAP-seq captured gRNA identity with an efficiency of 95%, as compared to 39% in the Perturb-seq setting or 89% in Perturb-seq with additional targeted gRNA amplification20”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [20] Hill, A. J. et al. On the design of CRISPR-based single-cell molecular screens. Nat. Methods 15, 271–274 (2018). source

      Not checked in the free run.

  • May combine two claims.

    Findingown data

    TAP-seq was used to generate dense enhancer-promoter interaction maps for 2.5% of the human genome, screening over 7,000 CRISPRi perturbations.

    “We applied TAP-seq to generate dense enhancer-promoter interaction maps within 2.5% of the human genome, screening over 7,000 distinct CRISPRi perturbations.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The protocol amplified target genes across cell types and in primary cells.

    “and the protocol amplified target genes irrespective of cell type and in primary cells.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    All 1,778 putatively active enhancers predicted from ENCODE data in two chromosome regions were perturbed, and effects on expressed protein-coding genes in the same regions were measured.

    “we perturbed all 1,778 putatively active enhancers predicted21 based on ENCODE data in two regions on chromosome 8 and 11, and identified effects on expressed protein-coding genes within the same regions”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [21] Zacher, B. et al. Accurate promoter and enhancer identification in 127 ENCODE and roadmap epigenomics cell types and tissues by GenoSTAN. PLoS One 12, e0169249 (2017). source

      Not checked in the free run.

    • Rests on an assumption: Enhancers mainly regulate genes located near them.

  • May combine two claims.

    Findingown data

    The absolute expression change needed for detection was about 0.03 UMI per cell, roughly one molecule per cell.

    “We quantified the absolute expression change required for an effect to be detected to be approximately 0.03 UMI/cell, corresponding to approximately one molecule per cell”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Each cell measured 68 target genes for chromosome 8 or 79 for chromosome 11, plus control enhancers.

    “Thus, in each cell, 68 (chr. 8) or 79 (chr. 11) target genes, plus control enhancers, were measured.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Hi-C interaction frequencies were significantly higher in strong and weak enhancer-target pairs and carried more information about functional interaction potential than linear distance.

    “interaction frequencies from Hi-C were significantly elevated in strong and weak ETPs (Figure 3f) and contained more information on the potential to form functional interactions than linear distance”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In 48% of cases (33% of strong pairs), another gene lay between the enhancer and target TSS; those intervening genes were on average expressed 1,000-fold lower than the true target.

    “However, in 48% of cases (33% of strong ETPs), at least one other gene was located between the enhancer and target TSS (Extended Data Figure 9f); these genes were on average expressed 1,000-fold lower compared to the true enhancer target (Extended Data Figure 9g).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Most enhancers, and nearly all from strong pairs, lie 10-50 kb from their target gene TSS.

    “most enhancers, and virtually all enhancers from strong ETPs, are located close (10-50kb) to their respective target gene’s transcriptional start site (TSS”
    Quote in the paperNot a substantial claim

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

    Findingown data

    In the whole transcriptome dataset, principal components 1-3 were not significantly associated with GO terms.

    “In the whole transcriptome dataset, PC1-3 were not significantly associated with GO-terms (not shown).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    A targeted readout of relevant genes could overcome limitations, permit genome-scale screens, and enable new applications.

    “a targeted readout of only relevant genes could overcome limitations, enable genome-scale screens, and open novel applications”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisereasoning

    A hypothetical genome-wide Perturb-seq screen would require 20,000 expression hypotheses per knockout, or 400 million tests in total.

    “For example in a hypothetical genome-wide Perturb-seq screen, 20,000 hypotheses on gene expression changes need to be tested for each of 20,000 knockouts (i.e. 400 million total tests).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Earlier studies limited differential expression testing to predefined candidate genes or grouped genes into signatures.

    “differential gene expression testing was restricted to pre-defined candidate genes7,9,10 or genes were grouped into gene signatures3,4,6,8.”
    Quote in the paperSubstantial claim0 of 7 sources checked
    • cites [7] Xie, S., Duan, J., Li, B., Zhou, P. & Hon, G. C. Multiplexed engineering and analysis of combinatorial enhancer activity in single cells. Mol. Cell 66, 285–299.e5 (2017). source

      Not checked in the free run.

    • cites [9] Gasperini, M. et al. A genome-wide framework for mapping gene regulation via cellular genetic screens. Cell 176, 377–390 (2019). source

      Not checked in the free run.

    • cites [10] Van Der Wijst, M. G. P. et al. Single-cell RNA sequencing identifies celltype-specific cis-eQTLs and co-expression QTLs. Nat. Genet. 50, 493–497 (2018). source

      Not checked in the free run.

    • cites [3] Shifrut, E. et al. Genome-wide CRISPR screens in primary human T cells reveal key regulators of immune function. Cell 175, 1958–1971.e15 (2018). source

      Not checked in the free run.

    • cites [4] Dixit, A. et al. Perturb-Seq: dissecting molecular circuits with scalable single-cell RNA profiling of pooled genetic screens. Cell 167, 1853–1866.e17 (2016). source

      Not checked in the free run.

    • cites [6] Datlinger, P. et al. Pooled CRISPR screening with single-cell transcriptome readout. Nat. Methods 14, 297–301 (2017). source

      Not checked in the free run.

    • cites [8] Adamson, B. et al. A multiplexed single-cell CRISPR screening platform enables systematic dissection of the unfolded protein response. Cell 167, 1867–1882.e21 (2016). source

      Not checked in the free run.

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

    Study design and conductown data

    sgRNA libraries were grown to at least 100-fold coverage.

    “grown out on plates for 16 h at 30 °C for ≥ 100x library coverage”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    By focusing coverage on selected genes, TAP-seq reduces sequencing needs up to 50-fold and addresses the multiple testing problem of whole transcriptome screens.

    “By focusing sequencing coverage on genes of interest, TAP-seq lowers sequencing requirements up to 50-fold, and solves the multiple hypothesis testing problem typically encountered in whole transcriptome screens.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Hypothesis-free analyses of single-cell genetic screens are dominated by false positives and cannot robustly identify weak hits.

    “These findings suggest that hypothesis-free analyses of single-cell genetic screens are strongly dominated by false positives and fail to robustly identify weak hits.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: True off-target effects are present and counted as false positives, so precision is underestimated.

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

    Premisecitation

    The L1000 panel could be used and has been shown to capture most transcriptome information across many perturbations.

    “Alternatively, gene panels such as the L1000 panel34 could be used (Extended Data Figure 3), which has been shown to capture the majority of information contained in the full transcriptome across a wide range of perturbations34.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [34] Subramanian, A. et al. A next generation connectivity Map: L1000 platform and the first 1,000,000 profiles. Cell 171, 1437–1452.e17 (2017). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    TAP-seq improves sensitivity to small expression changes and lowly expressed genes and recovers gRNA identity efficiently.

    “It increases the sensitivity towards small expression changes and lowly expressed genes, and retrieves the gRNA identity efficiently.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    TAP-seq's modular design provides platform-independent, robust genetic screens, unlike earlier targeted single-cell RNA-seq methods.

    “Unlike previous high-throughput27–29 and low-throughput30 targeted single-cell RNA-seq methods, its modular design ensures platform-independent, robust genetic screens.”
    Quote in the paperSubstantial claim0 of 4 sources checked
    • cites [27] Fan, H. C., Fu, G. K. & Fodor, S. P. A. Combinatorial labeling of single cells for gene expression cytometry. Science 347, 1258367 (2015). source

      Not checked in the free run.

    • cites [28] Shum, E. Y., Walczak, E. M., Chang, C. & Fan, H. C. in Advances in Experimental Medicine and Biology 63–79 (Springer, 2019). source

      Not checked in the free run.

    • cites [29] Vallejo, A. F.et al. Resolving cellular systems by ultra-sensitive and economical single-cell transcriptome filtering. Preprint at bioRxiv https://doi.org/10.1101/800631 (2019). source

      Not checked in the free run.

    • cites [30] Uzbas, F. et al. BART-Seq: cost-effective massively parallelized targeted sequencing for genomics, transcriptomics, and single-cell analysis. Genome Biol. 20, 155 (2019). source

      Not checked in the free run.

  • May combine two claims.

    Study design and conductown data

    The four enhancer-targeting guides were validated by pooled transduction into K562 dCas9-KRAB cells, with qPCR quantifying effects on target gene expression; HBE1 served as the target for the HS2 enhancer in three replicates.

    “Enhancer gRNAs were validated by pooled transduction of K562 dCas9-KRAB cells with all four enhancer-targeting guides, and the effect on target gene expression was quantified by qPCR. HBE1 was analyzed as target gene for the HS2 enhancer. n=3 replicates.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    A prior Perturb-seq study found enhancers for only about 3.5-6% of expressed genes, indicating TAP-seq is more sensitive.

    “A previous study using Perturb-seq identified enhancers for only ~3.5-6% of expressed genes9, underscoring the higher sensitivity of TAP-seq.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [9] Gasperini, M. et al. A genome-wide framework for mapping gene regulation via cellular genetic screens. Cell 176, 377–390 (2019). source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Study design and conductown data

    FACS selected single, viable cells with intermediate-to-high dCas9-BFP-KRAB expression.

    “FACS was carried out on a BD FACSAria Fusion flow cytometer (BD Biosciences) selecting single cells using forward and sideward scatter, viable cells using Draq7 dye exclusion and dCas9-BFP-KRAB intermediate to high expressing cells using BFP signal.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Enhancers within 1 kb of the target TSS were excluded.

    “also, enhancers within 1kb of the target TSS were excluded.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given

    For complex transcriptomic signatures, informative gene sets can be defined in advance from a whole transcriptome reference.

    “When applied to more complex transcriptomic signatures, e.g. of cellular differentiation/state, or immune cell activation, informative sets of genes can be defined a priori using a whole transcriptome reference.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    K562 CRISPRi cells were infected with CROP-seq vectors at low MOI to obtain mostly single infections.

    “Infection of K562 CRISPRi with CROP-seq vectors was done at low MOI to get mostly single infected cells.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationown data

    These results highlight the need for hypothesis-driven analysis in Perturb-seq and related designs.

    “These results emphasize the need for hypothesis-driven analysis in Perturb-seq and related experimental designs.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: True off-target effects are present and counted as false positives, so precision is underestimated.

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

    Study design and conductown data

    More than 20 colonies per library were picked and Sanger validated.

    “In parallel, > 20 colonies were picked per library and Sanger validated.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Earlier Perturb-seq enhancer-target studies were limited to large effects and found enhancers for only a small fraction of candidate genes (3.5-6% versus 24-32% here).

    “Previous applications of Perturb-seq for enhancer targets7,9 had remained limited to large effect sizes and identified enhancers only for a small fraction of candidate genes (3.5-6% in ref. 9 compared to 24-32% reported here).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [7] Xie, S., Duan, J., Li, B., Zhou, P. & Hon, G. C. Multiplexed engineering and analysis of combinatorial enhancer activity in single cells. Mol. Cell 66, 285–299.e5 (2017). source

      Not checked in the free run.

    • cites [9] Gasperini, M. et al. A genome-wide framework for mapping gene regulation via cellular genetic screens. Cell 176, 377–390 (2019). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    TAP-seq detects relative expression changes sensitively, and these measures do not depend on absolute expression detection accuracy.

    “TAP-seq allows highly sensitive detection of relative gene expression changes, and such measures do not depend on absolute expression level detection-accuracy.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    TAP-seq outperforms non-targeted methods for cell type mapping and reliably identifies cell types and differentiation states at very low sequencing depth.

    “In summary, TAP-seq outperforms non-targeted approaches in cell type mapping, and reliably identifies cell types and differentiation states at an extremely low sequencing depth.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The deeply sequenced whole transcriptome reference provides correct ground-truth cell type labels.

  • May combine two claims.

    Interpretationown data

    The targeted PCR captures target genes efficiently without reducing library complexity.

    “Together, these analyses show that the targeted PCR efficiently captures target genes without leading to a decrease in library complexity.”
    Quote in the paperSubstantial claim
45 of 45

  • The paper relies on this without support.

    Enhancers mainly regulate genes located near them.

    “Since enhancers predominantly affect genes in their proximity, assigning enhancer-target gene pairs (ETPs) is a well-suited task for TAP-seq.”

  • The paper relies on this without support.

    True off-target effects are present and counted as false positives, so precision is underestimated.

    “Of note, precision is underestimated in these analyses, since true off-target effects are classified as false positives.”

  • The paper relies on this without support.

    The deeply sequenced whole transcriptome reference provides correct ground-truth cell type labels.

    “compared to ground truth labels obtained from our deeply sequenced reference.”

  • The authors state this as an assumption.

    For precision-recall evaluation, the intended gRNA targets are treated as the complete set of true positives.

    “Precision-Recall curves were computed assuming that the intended gRNA targets constitute the full set of true positives.”
4 of 4

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?

  • The closest existing analysis for assessing calibration underestimates precision because downstream effects of perturbations are not taken into account.

    “this approach underestimates precision because downstream effects of perturbations are not taken into account”

    From Robust differential expression testing for single-cell CRISPR screens at low multiplicity of infection (2024) · cites this paper as [15]

  • There does not exist a standard protocol for assessing the calibration of single-cell CRISPR screen association methods.

    “there does not exist a standard protocol for assessing the calibration of single-cell CRISPR screen association methods”

    From Robust differential expression testing for single-cell CRISPR screens at low multiplicity of infection (2024) · cites this paper as [15]

    The Results section describes precision–recall analyses, and Extended Data Fig. 7 compares differential-expression testing methods, but neither addresses whether a standard protocol exists for assessing the calibration of single-cell CRISPR screen association methods. Searches for calibration, standardized assessment, and association-method benchmarking found no passage making that claim; page numbers were unavailable in our copy. (LLM, conf=0.76)

    Experimental — accuracy not yet measured

  • Tap-seq is a sensitive, low-cost, and platform-independent technique targeting specific genes for scRNA-seq coverage.

    “Moreover, Tap-seq [13] is a sensitive, low-cost, and platform-independent technique targeting specific genes for scRNA-seq coverage.”

    From A mini-review on perturbation modelling across single-cell omic modalities (2024) · cites this paper as [13]

  • Perturb-seq facilitates the analysis of thousands of CRISPR-mediated gene perturbations at single-cell resolution and can also be applied to common and rare cell populations.

    “Perturb-seq facilitates the analysis of thousands of CRISPR-mediated gene perturbations at single-cell resolution and can also be applied to common and rare cell populations.170”

    From Understanding the cellular interactome of non-alcoholic fatty liver disease (2022) · cites this paper as [170]

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Schraivogel, D., Gschwind, A. R., Milbank, J. H., Leonce, D. R., Jakob, P., Mathur, L., Korbel, J. O., Merten, C. A., Velten, L., & Steinmetz, L. M.
Affiliations
  • European Molecular Biology Laboratory (DE)
  • Stanford Medicine (US)
  • Barcelona Institute of Science and Technology (ES)
Corresponding author
  • Velten, Lars
  • Steinmetz, Lars M.
ORCID
Venue
Nature Methods
Publisher
Springer Science and Business Media LLC
Published
2020-06
Volume
17
Issue
6
Pages
629-635
ISSN
  • 1548-7091 (ISSN-L)
  • 1548-7105
DOI
10.1038/s41592-020-0837-5
PMC
PMC7610614
OpenAlex
W3029730770
Cited by
245
References
58
Citations by year
2026: 43 · 2025: 45 · 2024: 41 · 2023: 43 · 2022: 38 · 2021: 28 · 2020: 6 · 2013: 1
FWCI
9.40
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Methods (publishedVersion)
  • PubMed (publishedVersion)
  • PubMed Central (acceptedVersion)
  • Infoscience (Ecole Polytechnique Fédérale de Lausanne) (submittedVersion, cc-by-nc-nd)
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 Chromatin Dynamics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › CRISPR and Genetic Engineering
Keywords
single-cell RNA sequencing, epigenetic states, low-expression genes, sequencing sensitivity
MeSH terms
RNA-Seq, Humans, Genome, Human, Single-Cell Analysis, Transcriptome, Clustered Regularly Interspaced Short Palindromic Repeats

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 2026-10-04.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-04.

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