Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 47 of the 47 references it cites were found in a registry or an index.

RENGE infers gene regulatory networks using time-series single-cell RNA-seq data with CRISPR perturbations

Open access

Abstract

Single-cell RNA-seq analysis coupled with CRISPR-based perturbation has enabled the inference of gene regulatory networks with causal relationships. However, a snapshot of single-cell CRISPR data may not lead to an accurate inference, since a gene knockout can influence multi-layered downstream over time. Here, we developed RENGE, a computational method that infers gene regulatory networks using a time-series single-cell CRISPR dataset. RENGE models the propagation process of the effects elicited by a gene knockout on its regulatory network. It can distinguish between direct and indirect regulations, which allows for the inference of regulations by genes that are not knocked out. RENGE therefore outperforms current methods in the accuracy of inferring gene regulatory networks. When used on a dataset we derived from human-induced pluripotent stem cells, RENGE yielded a network consistent with multiple databases and literature. Accurate inference of gene regulatory networks by RENGE would enable the identification of key factors for various biological systems.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s42003-023-05594-4

Cites

47 of 47 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Overview of RENGE. a The use of snapshot expression data alone after gene X is knocked out makes it difficult to distinguish direct and indirect regulations. Blue nodes : genes with expression changes due to the KO. b Using time-series expression data after KO, in principle, can enable differentiation of direct and indirect regulations and infer regulation by non-KO genes. c RENGE infers a signed GRN from the time-series expression data after gene KO, obtained by scCRISPR analysis, by modeling the process in which the effects of the gene KO propagate on the network. X denotes the decrease in expression of the target gene due to KO (orange node), AX denotes the expression change due to direct regulation from the KO gene, and A2X denotes the expression change due to indirect regulation via another gene from the KO gene. w(t, k, g) denotes the magnitude of the effect of the k-th order regulation from the KO gene g at time t. d Comparison of methods on different features.
Fig. 1 Overview of RENGE. a The use of snapshot expression data alone after gene X is knocked out makes it difficult to distinguish direct and indirect regulations. Blue nodes : genes with expression changes due to the KO. b Using time-series expression data after KO, in principle, can enable differentiation of direct and indirect regulations and infer regulation by non-KO genes. c RENGE infers a signed GRN from the time-series expression data after gene KO, obtained by scCRISPR analysis, by modeling the process in which the effects of the gene KO propagate on the network. X denotes the decrease in expression of the target gene due to KO (orange node), AX denotes the expression change due to direct regulation from the KO gene, and A2X denotes the expression change due to indirect regulation via another gene from the KO gene. w(t, k, g) denotes the magnitude of the effect of the k-th order regulation from the KO gene g at time t. d Comparison of methods on different features.CC BY · PubMed Central
Fig. 2 Benchmark results when using the simulated data set. Horizontal axis: ratio of KO genes in the GRNs. log2(AUPRC ratio) were averaged between three backbones. a Regulation by KO genes. b Regulation by non-KO genes. c Positive regulation by KO genes. d Positive regulation by non-KO genes. e Negative regulation by KO genes. f Negative regulation by non-KO genes. Error bars represent mean ± standard deviation.
Fig. 2 Benchmark results when using the simulated data set. Horizontal axis: ratio of KO genes in the GRNs. log2(AUPRC ratio) were averaged between three backbones. a Regulation by KO genes. b Regulation by non-KO genes. c Positive regulation by KO genes. d Positive regulation by non-KO genes. e Negative regulation by KO genes. f Negative regulation by non-KO genes. Error bars represent mean ± standard deviation.CC BY · PubMed Central
Fig. 3 Time-series scCRISPR analysis of human iPS cells. a Experimental design. scRNA-seq was performed on cells sampled 2, 3, 4, and 5 days after transduction by lentiviruses containing a gRNA vector. b, c UMAP plot for cells bearing a single gRNA. d, e PCA plot for cells bearing the gRNA of the control or POU5F1. Colors indicate the sampling day of the cells (b, d) or the target gene of a gRNA detected in a cell (c, e). f Marker expression including UTF1 (pluripotency marker), SIX6 (neural markers), TBXT (axial mesoderm marker), and CST1 (definitive endoderm marker).
Fig. 3 Time-series scCRISPR analysis of human iPS cells. a Experimental design. scRNA-seq was performed on cells sampled 2, 3, 4, and 5 days after transduction by lentiviruses containing a gRNA vector. b, c UMAP plot for cells bearing a single gRNA. d, e PCA plot for cells bearing the gRNA of the control or POU5F1. Colors indicate the sampling day of the cells (b, d) or the target gene of a gRNA detected in a cell (c, e). f Marker expression including UTF1 (pluripotency marker), SIX6 (neural markers), TBXT (axial mesoderm marker), and CST1 (definitive endoderm marker).CC BY · PubMed Central
Fig. 4 Benchmark results for the hiPSC data. a AUPRC ratios for regulation by KO genes. Horizontal axis: threshold of the confidence level (−10×log10(MACS2q-value)) for the TF binding in ChIP-seq data. b AUPRC ratios for regulation by non-KO genes. c, d Precision comparison of the methods. Horizontal axis: number of inferred regulations corresponding to a threshold of a significance score such as p-value. c Regulations by KO genes. d Regulations by non-KO genes. e Schematic diagram showing the direct, indirect, and no path regulations. Solid black arrows: regulation of the ground-truth network. If these regulations are inferred, they are called direct. If the dashed arrows are inferred, they are called indirect regulation via the k − 1 (k ≥ 2) genes. If the gray arrow is inferred, it is called no path, which represents regulation that is neither direct nor indirect. f Classification of regulations inferred by each method. The ground-truth network was constructed with ChIP threshold = 300.
Fig. 4 Benchmark results for the hiPSC data. a AUPRC ratios for regulation by KO genes. Horizontal axis: threshold of the confidence level (−10×log10(MACS2q-value)) for the TF binding in ChIP-seq data. b AUPRC ratios for regulation by non-KO genes. c, d Precision comparison of the methods. Horizontal axis: number of inferred regulations corresponding to a threshold of a significance score such as p-value. c Regulations by KO genes. d Regulations by non-KO genes. e Schematic diagram showing the direct, indirect, and no path regulations. Solid black arrows: regulation of the ground-truth network. If these regulations are inferred, they are called direct. If the dashed arrows are inferred, they are called indirect regulation via the k − 1 (k ≥ 2) genes. If the gray arrow is inferred, it is called no path, which represents regulation that is neither direct nor indirect. f Classification of regulations inferred by each method. The ground-truth network was constructed with ChIP threshold = 300.CC BY · PubMed Central
Fig. 5 hiPSC regulatory network inferred by RENGE. a GRN of hiPSCs. Regulation was detected with FDR < 0.01. Only the edges of the top 35% absolute regulatory coefficients and the nodes with at least one edge are shown (474 edges and 95 nodes). Red arrows: positive regulation, blue arrows: negative regulation. Arrow thickness indicates the magnitude of the regulatory coefficient. Pink nodes: KO genes. b Number of positive and negative regulations from each gene. c The top 20 genes with the highest out-degree extracted from (b). The genes with asterisks indicate KO genes. d GRN of the top 20 genes with the highest out-degree. Only the edges of the top 20% absolute regulatory coefficients are shown. Solid arrows: regulations supported by ChIP-seq data (−10×log10(MACS2q-value)>50) obtained from ChIP-Atlas, and dotted arrows: regulations not supported. e The subnetwork at the center of the GRN in (d). All edges with FDR < 0.01 are shown. f Weighted out-degree of each gene in the GRN in (e). Weighted out-degree was calculated as the sum of absolute regulatory coefficients from each gene.
Fig. 5 hiPSC regulatory network inferred by RENGE. a GRN of hiPSCs. Regulation was detected with FDR < 0.01. Only the edges of the top 35% absolute regulatory coefficients and the nodes with at least one edge are shown (474 edges and 95 nodes). Red arrows: positive regulation, blue arrows: negative regulation. Arrow thickness indicates the magnitude of the regulatory coefficient. Pink nodes: KO genes. b Number of positive and negative regulations from each gene. c The top 20 genes with the highest out-degree extracted from (b). The genes with asterisks indicate KO genes. d GRN of the top 20 genes with the highest out-degree. Only the edges of the top 20% absolute regulatory coefficients are shown. Solid arrows: regulations supported by ChIP-seq data (−10×log10(MACS2q-value)>50) obtained from ChIP-Atlas, and dotted arrows: regulations not supported. e The subnetwork at the center of the GRN in (d). All edges with FDR < 0.01 are shown. f Weighted out-degree of each gene in the GRN in (e). Weighted out-degree was calculated as the sum of absolute regulatory coefficients from each gene.CC BY · PubMed Central
Fig. 6 Relationship between regulatory correlations and protein complex formation. Regulatory correlations are calculated as correlations between the regulatory-coefficient vectors for each gene pair in the network. A positive regulatory correlation indicates that the same genes are regulated in the same direction by the pair, while a negative regulatory correlation indicates that the same genes are regulated in the opposite direction. a Distribution of the regulatory correlations among all possible gene pairs versus those in the genes in protein complexes obtained from CORUM3.0 versus those among gene pairs with a high STRING score. The STRING score represents the confidence level of the protein-protein interactions. b Top 10 gene pairs with the highest absolute regulatory correlation among gene pairs with a positive STRING score. c Examples of gene pairs with large negative and positive regulatory correlations. d Relationship between regulatory correlations and colocalization scores of TF binding. Colocalization was calculated from ChIP-seq data in pluripotent stem cells by the ChIP-Atlas. Colors indicate STRING scores. Gene names for the top 3 pairs with the highest colocalization score are shown.
Fig. 6 Relationship between regulatory correlations and protein complex formation. Regulatory correlations are calculated as correlations between the regulatory-coefficient vectors for each gene pair in the network. A positive regulatory correlation indicates that the same genes are regulated in the same direction by the pair, while a negative regulatory correlation indicates that the same genes are regulated in the opposite direction. a Distribution of the regulatory correlations among all possible gene pairs versus those in the genes in protein complexes obtained from CORUM3.0 versus those among gene pairs with a high STRING score. The STRING score represents the confidence level of the protein-protein interactions. b Top 10 gene pairs with the highest absolute regulatory correlation among gene pairs with a positive STRING score. c Examples of gene pairs with large negative and positive regulatory correlations. d Relationship between regulatory correlations and colocalization scores of TF binding. Colocalization was calculated from ChIP-seq data in pluripotent stem cells by the ChIP-Atlas. Colors indicate STRING scores. Gene names for the top 3 pairs with the highest colocalization score are shown.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

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?

Self-check

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

The self-check needs the paper’s full text.

Citations

Cites

Show all 45 references

Cited by

See all 38

Metadata

Authors
Ishikawa, M., Sugino, S., Masuda, Y., Tarumoto, Y., Seto, Y., Taniyama, N., Wagai, F., Yamauchi, Y., Kojima, Y., Kiryu, H., Yusa, K., Eiraku, M., & Mochizuki, A.
Affiliations
  • Kyoto University (JP)
  • National Cancer Center Research Institute (JP)
  • The University of Tokyo (JP)
Corresponding author
Ishikawa, Masato
Venue
Communications Biology
Published
2023-12-28
Volume
6
Issue
1
Pages
1290-1290
ISSN
2399-3642 (ISSN-L)
OpenAlex
W4390340766
Cited by
22
References
45
Citations by year
2026: 10 · 2025: 9 · 2024: 3
FWCI
1.95
Type
journal article
Language
en
Open access
Licence
Locations
  • Communications Biology (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Institutional Repositories DataBase (IRDB) (submittedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Kyoto University Research Information Repository (Kyoto University) (submittedVersion, cc-by)
Funders
  • MEXT | Japan Society for the Promotion of Science (JP20K15714, JP22K06237, 17K00398, 20K12059, JP17K20146, JP21H04959, 19H03196)
  • Takeda Science Foundation
  • Kato Memorial Bioscience Foundation
  • Joint Usage/Research Center program of Institute for Life and Medical Sciences, Kyoto University
  • MEXT | JST | Core Research for Evolutional Science and Technology (JPMJCR1922)
  • Mitsubishi Foundation
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Gene Regulatory Network Analysis
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Bioinformatics and Genomic Networks
Keywords
gene regulatory network inference, CRISPR perturbation, gene knockout, causal inference, induced pluripotent stem cells
MeSH terms
Single-Cell Gene Expression Analysis, Humans, Time Factors, Gene Regulatory Networks, Gene Knockout Techniques

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.

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

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: RENGE infers gene regulatory networks using time-series single-cell RNA-seq data with CRISPR perturbations. https://app.cochat.ai/papers/10.1038/s42003-023-05594-4
The paper
Ishikawa, M., Sugino, S., Masuda, Y., Tarumoto, Y., Seto, Y., Taniyama, N., Wagai, F., Yamauchi, Y., Kojima, Y., Kiryu, H., Yusa, K., Eiraku, M., & Mochizuki, A. (2023). RENGE infers gene regulatory networks using time-series single-cell RNA-seq data with CRISPR perturbations. Communications Biology. https://doi.org/10.1038/s42003-023-05594-4

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.