CoChat

Gene knockout inference with variational graph autoencoder learning single-cell gene regulatory networks

Open access

The DOI matches its Crossref record; 70 of the 70 references it cites were found in a registry or an index.

Abstract

In this paper, we introduce Gene Knockout Inference (GenKI), a virtual knockout (KO) tool for gene function prediction using single-cell RNA sequencing (scRNA-seq) data in the absence of KO samples when only wild-type (WT) samples are available. Without using any information from real KO samples, GenKI is designed to capture shifting patterns in gene regulation caused by the KO perturbation in an unsupervised manner and provide a robust and scalable framework for gene function studies. To achieve this goal, GenKI adapts a variational graph autoencoder (VGAE) model to learn latent representations of genes and interactions between genes from the input WT scRNA-seq data and a derived single-cell gene regulatory network (scGRN). The virtual KO data is then generated by computationally removing all edges of the KO gene—the gene to be knocked out for functional study—from the scGRN. The differences between WT and virtual KO data are discerned by using their corresponding latent parameters derived from the trained VGAE model. Our simulations show that GenKI accurately approximates the perturbation profiles upon gene KO and outperforms the state-of-the-art under a series of evaluation conditions. Using publicly available scRNA-seq data sets, we demonstrate that GenKI recapitulates discoveries of real-animal KO experiments and accurately predicts cell type-specific functions of KO genes. Thus, GenKI provides an in-silico alternative to KO experiments that may partially replace the need for genetically modified animals or other genetically perturbed 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 · 10.1093/nar/gkad450

Cites

70 of 70 references verified0 citation errors · 0 not checkable

Statistics

Figures

Graphical Abstract
Graphical Abstract CC BY · PubMed Central
Figure 1. The pipeline contains seven steps: (1) construction of WT scGRN, (2) training VGAE model, (3) transfer the trained VGAE model, (4) construction of virtual KO data, (5) latent embeddings of WT and virtual KO data, (6) calculation of KL divergence, and (7) identification of KO-responsive genes for function annotation and analysis.
Figure 1. The pipeline contains seven steps: (1) construction of WT scGRN, (2) training VGAE model, (3) transfer the trained VGAE model, (4) construction of virtual KO data, (5) latent embeddings of WT and virtual KO data, (6) calculation of KL divergence, and (7) identification of KO-responsive genes for function annotation and analysis.CC BY · PubMed Central
Figure 2. Three methods were included in the comparison including scTenifoldKnk and two baseline predictors, which are based on random rankings and Pearson's correlation, respectively (Materials and Methods). (A) The procedure of assessment of virtual KO tools using simulated data sets. (B) The levels of AUROC of virtual KO experiments using three simulated SERGIO data sets. (C) The levels of AP of virtual KO experiments using three simulated SERGIO data sets. Size represents the number of genes in each data set.
Figure 2. Three methods were included in the comparison including scTenifoldKnk and two baseline predictors, which are based on random rankings and Pearson's correlation, respectively (Materials and Methods). (A) The procedure of assessment of virtual KO tools using simulated data sets. (B) The levels of AUROC of virtual KO experiments using three simulated SERGIO data sets. (C) The levels of AP of virtual KO experiments using three simulated SERGIO data sets. Size represents the number of genes in each data set.CC BY · PubMed Central
Figure 3. Trem2-KO responsive genes inferred by GenKI. (A) Illustration of Trem2-KO experiment generating the microglia data set. (B) GO terms significantly enriched in functions of Trem2-KO responsive genes. The –log10-transformed adjusted p-value indicates the strength of enrichment for each term. (C) STRING network of Trem2-KO responsive genes. Edge thickness indicates the strength of data support. (D) Adjusted R2 score of the regression of expression levels by setting the Trem2 as a response variable and other KO-responsive genes as explanatory variables, compared to that of randomly selected genes as the explanatory variables.
Figure 3. Trem2-KO responsive genes inferred by GenKI. (A) Illustration of Trem2-KO experiment generating the microglia data set. (B) GO terms significantly enriched in functions of Trem2-KO responsive genes. The –log10-transformed adjusted p-value indicates the strength of enrichment for each term. (C) STRING network of Trem2-KO responsive genes. Edge thickness indicates the strength of data support. (D) Adjusted R2 score of the regression of expression levels by setting the Trem2 as a response variable and other KO-responsive genes as explanatory variables, compared to that of randomly selected genes as the explanatory variables.CC BY · PubMed Central
Figure 4. Nkx2-1-KO responsive genes inferred by GenKI. (A) GO terms significantly enriched in functions of Nkx2-1-KO responsive genes. The -log10-transformed adjusted p-value indicates the strength of enrichment for each term. (B) STRING network consists of Nkx2-1-KO responsive genes. The zoomed inset demonstrates a subnetwork module containing the KO gene.
Figure 4. Nkx2-1-KO responsive genes inferred by GenKI. (A) GO terms significantly enriched in functions of Nkx2-1-KO responsive genes. The -log10-transformed adjusted p-value indicates the strength of enrichment for each term. (B) STRING network consists of Nkx2-1-KO responsive genes. The zoomed inset demonstrates a subnetwork module containing the KO gene.CC BY · PubMed Central
Figure 5. Hnf4a & Smad4-KO responsive genes inferred by GenKI. (A) GO terms significantly enriched in functions of inferred Hnf4a-Smad4-KO responsive genes. The –log10-transformed p-value indicates the strength of enrichment for each term. (B) STRING subnetwork consists of Hnf4a-Smad4-KO responsive genes.
Figure 5. Hnf4a & Smad4-KO responsive genes inferred by GenKI. (A) GO terms significantly enriched in functions of inferred Hnf4a-Smad4-KO responsive genes. The –log10-transformed p-value indicates the strength of enrichment for each term. (B) STRING subnetwork consists of Hnf4a-Smad4-KO responsive genes.CC BY · PubMed Central
Figure 6. Venn diagrams and barcode enrichment plots showing the intersection and differences between the KO responsive genes given by GenKI and DE genes. Venn diagram and barcode enrichment plot of (A) microglia data set, (B) lung data set and (C) intestine data set. All the numbers of overlapped genes were significantly greater than random expectations (p-value < 10E-05, hypergeometric test).
Figure 6. Venn diagrams and barcode enrichment plots showing the intersection and differences between the KO responsive genes given by GenKI and DE genes. Venn diagram and barcode enrichment plot of (A) microglia data set, (B) lung data set and (C) intestine data set. All the numbers of overlapped genes were significantly greater than random expectations (p-value < 10E-05, hypergeometric test).CC BY · PubMed Central
Figure 7. STAT1-KO responsive genes inferred by GenKI. (A) GO terms significantly enriched in functions of STAT1-KO responsive genes. The –log10-transformed adjusted p-value indicates the strength of enrichment for each term. (B) STRING network consists of STAT1-KO responsive genes.
Figure 7. STAT1-KO responsive genes inferred by GenKI. (A) GO terms significantly enriched in functions of STAT1-KO responsive genes. The –log10-transformed adjusted p-value indicates the strength of enrichment for each term. (B) STRING network consists of STAT1-KO responsive genes.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 70 references

Research works printed but not linked: 1

  1. Glorot, 2010, Proceedings of the thirteenth international conference on artificial intelligence and statistics, 249

Cited by

See all 57

Metadata

Authors
Yang, Y., Li, G., Zhong, Y., Xu, Q., Chen, B.-J., Lin, Y.-T., Chapkin, R. S., & Cai, J. J.
Affiliations
  • Texas A&M University (US)
  • East China Normal University (CN)
  • National Chung Hsing University (TW)
  • National Taiwan University (TW)
Corresponding author
Yang, Yongjian
Venue
Nucleic Acids Research
Published
2023-05-29
Volume
51
Issue
13
Pages
6578-6592
ISSN
  • 0305-1048 (ISSN-L)
  • 1362-4962
OpenAlex
W4378640029
Cited by
48
References
69
Citations by year
2026: 22 · 2025: 15 · 2024: 10 · 2023: 1
FWCI
4.37
Type
journal article
Language
en
Open access
Locations
  • Nucleic Acids Research (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
Funders
  • National Institute of Environmental Health Sciences (P30 ES029067)
  • National Cancer Institute (R35 CA197707, RO1 CA245514)
  • Allen Endowed Chair in Nutrition & Chronic Disease Prevention
  • U.S. Department of Defense (GW200026)
  • Texas A&M University
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
virtual knockout, gene function prediction, variational graph autoencoder, single-cell RNA sequencing, unsupervised learning, gene perturbation
MeSH terms
Animals, Gene Expression Regulation, Sequence Analysis, RNA, Gene Expression Profiling, Gene Regulatory Networks, Gene Knockout Techniques, 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.

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: Gene knockout inference with variational graph autoencoder learning single-cell gene regulatory networks. https://app.cochat.ai/papers/10.1093/nar/gkad450
The paper
Yang, Y., Li, G., Zhong, Y., Xu, Q., Chen, B.-J., Lin, Y.-T., Chapkin, R. S., & Cai, J. J. (2023). Gene knockout inference with variational graph autoencoder learning single-cell gene regulatory networks. Nucleic Acids Research. https://doi.org/10.1093/nar/gkad450

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.