CoChat

Chromatin accessibility dynamics of myogenesis at single cell resolution

Open access

The DOI matches its Crossref record; 49 of the 49 references it cites were found in a registry or an index.

Abstract

Over a million DNA regulatory elements have been cataloged in the human genome, but linking these elements to the genes that they regulate remains challenging. We introduce Cicero, a statistical method that connects regulatory elements to target genes using single cell chromatin accessibility data. We apply Cicero to investigate how thousands of dynamically accessible elements orchestrate gene regulation in differentiating myoblasts. Groups of co-accessible regulatory elements linked by Cicero meet criteria of “chromatin hubs”, in that they are physically proximal, interact with a common set of transcription factors, and undergo coordinated changes in histone marks that are predictive of gene expression. Pseudotemporal analysis revealed a subset of elements bound by MYOD in myoblasts that exhibit early opening, potentially serving as the initial sites of recruitment of chromatin remodeling and histone-modifying enzymes. The methodological framework described here constitutes a powerful new approach for elucidating the architecture, grammar and mechanisms of cis- regulation on a genome-wide basis.

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 · 10.1101/155473

Cites

49 of 49 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 Differentiating myoblasts follow similar single cell chromatin accessibility and gene expression trajectories. A ) Single cell chromatin accessibility profiles for human skeletal muscle myoblasts (HSMM) were constructed with sci-ATAC-seq. Contaminating interstitial fibroblasts (common in HSMM cultures) were removed informatically prior to further analysis. B ) The single cell trajectory inferred from 2,007 myoblast sci-ATAC-seq profiles by Monocle (see Methods). In subsequent panels and throughout the paper, we exclude cells on the branch to outcome F2 unless otherwise indicated, leaving 1,797 cells on the root and F1 branches. Inset shows the sc-RNA-seq trajectory reported for HSMMs (reproduced from Figure 2 of Qiu et al. (2017)). C ) Distribution of cells in chromatin accessibility pseudotime from the root to trajectory outcome F1. D ) Percent of differentiating cells whose promoters for selected genes are accessible across pseudotime. Black line indicates the pseudotime-dependent average from a smoothed binomial regression. E ) Percent of cells whose promoters for selected genes in D are accessible in fibroblasts collected in growth medium (GM) or differentiation medium (DM), as well as myoblasts localized to the branch to F2. F ) Change in promoter accessibility versus change in gene expression. Log2 fold changes were calculated by dividing the pseudotime-dependent average from a smoothed negative binomial regression at the 90th percentile of pseudotime by the average at the 10th percentile of pseudotime. A pseudocount of 0.1 was added to each prior to computing the fold changes.
Figure 1 Differentiating myoblasts follow similar single cell chromatin accessibility and gene expression trajectories. A ) Single cell chromatin accessibility profiles for human skeletal muscle myoblasts (HSMM) were constructed with sci-ATAC-seq. Contaminating interstitial fibroblasts (common in HSMM cultures) were removed informatically prior to further analysis. B ) The single cell trajectory inferred from 2,007 myoblast sci-ATAC-seq profiles by Monocle (see Methods). In subsequent panels and throughout the paper, we exclude cells on the branch to outcome F2 unless otherwise indicated, leaving 1,797 cells on the root and F1 branches. Inset shows the sc-RNA-seq trajectory reported for HSMMs (reproduced from Figure 2 of Qiu et al. (2017)). C ) Distribution of cells in chromatin accessibility pseudotime from the root to trajectory outcome F1. D ) Percent of differentiating cells whose promoters for selected genes are accessible across pseudotime. Black line indicates the pseudotime-dependent average from a smoothed binomial regression. E ) Percent of cells whose promoters for selected genes in D are accessible in fibroblasts collected in growth medium (GM) or differentiation medium (DM), as well as myoblasts localized to the branch to F2. F ) Change in promoter accessibility versus change in gene expression. Log2 fold changes were calculated by dividing the pseudotime-dependent average from a smoothed negative binomial regression at the 90th percentile of pseudotime by the average at the 10th percentile of pseudotime. A pseudocount of 0.1 was added to each prior to computing the fold changes.CC-BY-ND · From the PDF
Supplemental Figure 1. A ) Overlap between ENCODE DnaseI hypersensitive sites (DHSs) and MACS-called sciATAC-seq peaks visualized as an UpSetR plot (Lex and Gehlenborg, 2014). Horizontal bars indicate the number of peaks in each data set. Vertical bars indicate the number of peaks in each compartment of the three-way Venn diagram. Black dots indicate which datasets are involved (i.e. intersected) in each vertical bar. B ) Histogram of the number of cells with a read in each sci-ATAC-seq site, faceted by whether the site overlapped ENCODE DHS peaks in myoblasts or myotubes. C) Boxplot of the number of MACS-called sci-ATAC-seq peaks per cell. Promoter-proximal peaks are peaks intersecting the first 500 base pairs upstream of a transcription start site (see Methods). Distal peaks are all other peaks.
Supplemental Figure 1. A ) Overlap between ENCODE DnaseI hypersensitive sites (DHSs) and MACS-called sciATAC-seq peaks visualized as an UpSetR plot (Lex and Gehlenborg, 2014). Horizontal bars indicate the number of peaks in each data set. Vertical bars indicate the number of peaks in each compartment of the three-way Venn diagram. Black dots indicate which datasets are involved (i.e. intersected) in each vertical bar. B ) Histogram of the number of cells with a read in each sci-ATAC-seq site, faceted by whether the site overlapped ENCODE DHS peaks in myoblasts or myotubes. C) Boxplot of the number of MACS-called sci-ATAC-seq peaks per cell. Promoter-proximal peaks are peaks intersecting the first 500 base pairs upstream of a transcription start site (see Methods). Distal peaks are all other peaks.CC-BY-ND · From the PDF
Image 3
CC-BY-ND · From the PDF
Supplemental Figure 2. A ) Gene set enrichment analysis of significantly opening and closing accessible sites. Adjusted p-values were computed using a hypergeometric test. Terms shown are all sites with an adjusted p-value < 1e-4 in either the opening set or the closing set. Color represent the -log10 adjusted p-value. Sites are ordered by the -log10 adjusted p-value of the opening set. B ) Smoothed pseudotime-dependent accessibility curves, generated by a negative binomial regression of each for a set of selected cell cycle relevant genes. Each row indicates a different DNA element. Annotation column represents the -log10 adjusted p-value for the test of differential accessibility across pseudotime. For visualization, fitted curve range was capped at 100. C ) Percent of dynamic and static sites with changing Segway state assignment from myoblast to myotube.
Supplemental Figure 2. A ) Gene set enrichment analysis of significantly opening and closing accessible sites. Adjusted p-values were computed using a hypergeometric test. Terms shown are all sites with an adjusted p-value < 1e-4 in either the opening set or the closing set. Color represent the -log10 adjusted p-value. Sites are ordered by the -log10 adjusted p-value of the opening set. B ) Smoothed pseudotime-dependent accessibility curves, generated by a negative binomial regression of each for a set of selected cell cycle relevant genes. Each row indicates a different DNA element. Annotation column represents the -log10 adjusted p-value for the test of differential accessibility across pseudotime. For visualization, fitted curve range was capped at 100. C ) Percent of dynamic and static sites with changing Segway state assignment from myoblast to myotube.CC-BY-ND · From the PDF
Figure 3 Cicero constructs cis-regulatory models genome-wide from sci-ATAC-seq data. A ) An overview of the Cicero algorithm (see Methods for details) B ) Fraction of pairs of sites with positive co-accessibility scores at varying distances in the linear genome. C ) The mean number of distal sites per promoter and promoters per distal site at increasing co-accessibility score cutoffs. D ) A summary of the Cicero co-accessibility network containing the ID1 locus. Connections to elements outside the window shown are excluded for clarity. The height of connections indicates the magnitude of the Cicero co-accessibility score between the connected peaks. The conservation track shows the 100 vertebrate PhyloP conservation scores (phyloP100wayAll). The DNase-seq track shows the ENCODE signal track for HSMM. E ) Number of CCANs with 3 or more co-accessible DNA elements identified with Louvain community detection. Prior to running Louvain, connections below the indicated Cicero co-accessibility score were excluded (see Methods for details). CCANs that include the promoter of at least 1 detectably expressed gene are shown as a separate series. F ) Boxplots of the number of expressed gene promoters per CCAN at increasing coaccessibility score cutoffs. G ) Percent of sites recruited into a CCAN at increasing co-accessibility score cutoff. Colors represent subsets of sites: green represent promoters for genes that are detectably expressed; orange and red represent sites that are accessible and differentially accessible across pseudotime, respectively.
Figure 3 Cicero constructs cis-regulatory models genome-wide from sci-ATAC-seq data. A ) An overview of the Cicero algorithm (see Methods for details) B ) Fraction of pairs of sites with positive co-accessibility scores at varying distances in the linear genome. C ) The mean number of distal sites per promoter and promoters per distal site at increasing co-accessibility score cutoffs. D ) A summary of the Cicero co-accessibility network containing the ID1 locus. Connections to elements outside the window shown are excluded for clarity. The height of connections indicates the magnitude of the Cicero co-accessibility score between the connected peaks. The conservation track shows the 100 vertebrate PhyloP conservation scores (phyloP100wayAll). The DNase-seq track shows the ENCODE signal track for HSMM. E ) Number of CCANs with 3 or more co-accessible DNA elements identified with Louvain community detection. Prior to running Louvain, connections below the indicated Cicero co-accessibility score were excluded (see Methods for details). CCANs that include the promoter of at least 1 detectably expressed gene are shown as a separate series. F ) Boxplots of the number of expressed gene promoters per CCAN at increasing coaccessibility score cutoffs. G ) Percent of sites recruited into a CCAN at increasing co-accessibility score cutoff. Colors represent subsets of sites: green represent promoters for genes that are detectably expressed; orange and red represent sites that are accessible and differentially accessible across pseudotime, respectively.CC-BY-ND · From the PDF
Figure 4 Co-accessible DNA elements linked by Cicero are physically proximal in the nucleus. A ) Observed contact frequencies for pairs of sites linked by Cicero at varying distances in the linear genome, normalized to expected contact frequencies. Sites not connected by Cicero are shown for comparison. 'Whole genome' indicates the observed:expected ratio for all pairs of genomic bins. B ) Cicero connections for the LYN locus compared to RNA polII ChIA-PET data. C ) CCAN connections detected in ChIA-PET contacts as a function of distance. Color indicates the magnitude of the Cicero co-accessibility score. D ) ChIA-PET contacts found in Cicero connections as a function of distance. Colors in D and E indicate the magnitude of the Cicero co-accessibility scores included. E ) ChIA-PET contacts detected in Cicero CCAN connections as a function of ChIA-PET score.
Figure 4 Co-accessible DNA elements linked by Cicero are physically proximal in the nucleus. A ) Observed contact frequencies for pairs of sites linked by Cicero at varying distances in the linear genome, normalized to expected contact frequencies. Sites not connected by Cicero are shown for comparison. 'Whole genome' indicates the observed:expected ratio for all pairs of genomic bins. B ) Cicero connections for the LYN locus compared to RNA polII ChIA-PET data. C ) CCAN connections detected in ChIA-PET contacts as a function of distance. Color indicates the magnitude of the Cicero co-accessibility score. D ) ChIA-PET contacts found in Cicero connections as a function of distance. Colors in D and E indicate the magnitude of the Cicero co-accessibility scores included. E ) ChIA-PET contacts detected in Cicero CCAN connections as a function of ChIA-PET score.CC-BY-ND · From the PDF
Supplemental Figure 3 . A ) Percent of pol II ChIA-PET anchors within 1 kb of an sci-ATAC-seq peak as a function of ChIA-PET anchor score provided by Tang et. al. (2015). B ) Percent of sci-ATAC-seq peaks within 1 kb of pol II ChIAPET anchors as a function of overall cell accessibility (number of cells where the peak is accessible).
Supplemental Figure 3 . A ) Percent of pol II ChIA-PET anchors within 1 kb of an sci-ATAC-seq peak as a function of ChIA-PET anchor score provided by Tang et. al. (2015). B ) Percent of sci-ATAC-seq peaks within 1 kb of pol II ChIAPET anchors as a function of overall cell accessibility (number of cells where the peak is accessible).CC-BY-ND · From the PDF
Supplemental Figure 4. Motifs in accessible sites predict motif content of Cicero-linked sites. The network summarizes a graphical model that captures how occurrences of motifs in pairs of sites predict whether they are connected. Each motif is connected to the motifs it suggests will exist in one or more connected sites. A motif that predicts itself in a connected site is shown in dark blue. If motif 'A' at a distal site predicts that 'B' will be found at a promoter, and symmetrically 'B' at a distal site suggests 'A' will be found at a promoter, they are connected with a black line, with a width proportional to the strength of the co-accessibility. Asymmetric motif relationships are not shown.
Supplemental Figure 4. Motifs in accessible sites predict motif content of Cicero-linked sites. The network summarizes a graphical model that captures how occurrences of motifs in pairs of sites predict whether they are connected. Each motif is connected to the motifs it suggests will exist in one or more connected sites. A motif that predicts itself in a connected site is shown in dark blue. If motif 'A' at a distal site predicts that 'B' will be found at a promoter, and symmetrically 'B' at a distal site suggests 'A' will be found at a promoter, they are connected with a black line, with a width proportional to the strength of the co-accessibility. Asymmetric motif relationships are not shown.CC-BY-ND · From the PDF
Figure 5 Co-accessible DNA elements linked by Cicero are epigenetically co-modified. A ) Odds ratio of a site gaining H3K27ac during myoblast differentiation, given that it is linked to a site that is doing so. Color indicates the strength of the Cicero co-accessibility links. The lightest color indicates pairs of sites that are unlinked by Cicero. B ) Correspondence between a statically accessible site's gain or loss of H3K27ac and its maximum co-accessibility score to a site that is opening (x axis) or closing (y axis). Sites that are not linked to an opening or closing site are drawn at x = 0 or y = 0, respectively. C ) Similar to panel B, but describing the correspondence between a site's gain or loss of H3K27ac and its maximum co-accessibility score to a site that is gaining or losing MYOD. D ) The variance explained in a series of linear regression models in which the (Gaussian) response is the log2 fold change in H3K27ac level of each DNA element and the predictors are whether that site is opening, closing, or static, whether it undergoes gains or loses in MYOD binding, and whether it is linked to neighbors that are doing so. See Methods for full details on model specifications. E ) The Cicero map for the 755kb region surrounding MYH3 along with called MYOD ChIP-seq peaks from (Cao et al., 2010). Sites opening in accessibility are colored by their opening pseudotime (see Methods), sites that are not opening in accessibility are shown in grey. The inset shows the 60kb region surrounding MYH3 along with MYOD ChIP-seq and H3K27ac ChIP-seq signal tracks from (Cao et al., 2010) and (The ENCODE Project Consortium, 2012). F ) Opening pseudotimes for all opening sites, subdivided by whether
Figure 5 Co-accessible DNA elements linked by Cicero are epigenetically co-modified. A ) Odds ratio of a site gaining H3K27ac during myoblast differentiation, given that it is linked to a site that is doing so. Color indicates the strength of the Cicero co-accessibility links. The lightest color indicates pairs of sites that are unlinked by Cicero. B ) Correspondence between a statically accessible site's gain or loss of H3K27ac and its maximum co-accessibility score to a site that is opening (x axis) or closing (y axis). Sites that are not linked to an opening or closing site are drawn at x = 0 or y = 0, respectively. C ) Similar to panel B, but describing the correspondence between a site's gain or loss of H3K27ac and its maximum co-accessibility score to a site that is gaining or losing MYOD. D ) The variance explained in a series of linear regression models in which the (Gaussian) response is the log2 fold change in H3K27ac level of each DNA element and the predictors are whether that site is opening, closing, or static, whether it undergoes gains or loses in MYOD binding, and whether it is linked to neighbors that are doing so. See Methods for full details on model specifications. E ) The Cicero map for the 755kb region surrounding MYH3 along with called MYOD ChIP-seq peaks from (Cao et al., 2010). Sites opening in accessibility are colored by their opening pseudotime (see Methods), sites that are not opening in accessibility are shown in grey. The inset shows the 60kb region surrounding MYH3 along with MYOD ChIP-seq and H3K27ac ChIP-seq signal tracks from (Cao et al., 2010) and (The ENCODE Project Consortium, 2012). F ) Opening pseudotimes for all opening sites, subdivided by whetherCC-BY-ND · From the PDF
Image 10
CC-BY-ND · From the PDF
Image 11
CC-BY-ND · From the PDF
Figure 6 Chromatin dynamics at distal DNA elements predicts gene regulation. A ) Changes in histone acetylation in the first 1 kb downstream of each gene's TSS, corresponding to the 'barrier' to RNA pol II elongation posed by nucleosomes, are correlated with changes in its expression. B ) Two regression models predict changes in the histone marks deposited throughout each gene's barrier region. The first model predicts changes on the basis of transcription factor binding motifs in gene promoters. The second model adds variables encoding the strength of co-accessibility with linked sites containing the motif. See Methods for details on the various models. Adjusted R^2 is computed as the fraction of null deviance explained. The number to the right of each bar indicates the ratio of variance explained between the first and second model. C ) Similar to panel B, with changes in expression as the response. D ) Coefficients from the model incorporating sequence at distal sites for each motif surviving model selection via elastic net. Note that the model considers each motif twice: once at promoters and again at distal sites, and both can be selected by elastic net.
Figure 6 Chromatin dynamics at distal DNA elements predicts gene regulation. A ) Changes in histone acetylation in the first 1 kb downstream of each gene's TSS, corresponding to the 'barrier' to RNA pol II elongation posed by nucleosomes, are correlated with changes in its expression. B ) Two regression models predict changes in the histone marks deposited throughout each gene's barrier region. The first model predicts changes on the basis of transcription factor binding motifs in gene promoters. The second model adds variables encoding the strength of co-accessibility with linked sites containing the motif. See Methods for details on the various models. Adjusted R^2 is computed as the fraction of null deviance explained. The number to the right of each bar indicates the ratio of variance explained between the first and second model. C ) Similar to panel B, with changes in expression as the response. D ) Coefficients from the model incorporating sequence at distal sites for each motif surviving model selection via elastic net. Note that the model considers each motif twice: once at promoters and again at distal sites, and both can be selected by elastic net.CC-BY-ND · From the PDF

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 49 references

Cited by

See all 16

Metadata

Authors
Pliner, H. A., Packer, J., McFaline-Figueroa, J. L., Cusanovich, D. A., Daza, R., Srivatsan, S., Qiu, X., Jackson, D., Minkina, A., Adey, A., Steemers, F. J., Shendure, J., & Trapnell, C.
Affiliations
  • University of Washington (US)
  • Oregon Health & Science University (US)
  • Illumina (United States) (US)
  • Howard Hughes Medical Institute (US)
Venue
bioRxiv (Cold Spring Harbor Laboratory)
Publisher
openRxiv
Published
2017-06-26
OpenAlex
W2653530754
Cited by
16
References
48
Citations by year
2021: 1 · 2020: 2 · 2019: 5 · 2018: 7 · 2017: 1
Type
posted content
Language
en
Open access
Locations
bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by-nd)
Funders
  • National Science Foundation (T32HL007828, 1256082, 5T32HG000035)
  • Howard Hughes Medical Institute
  • Alfred P. Sloan Foundation
  • Paul G. Allen Family Foundation
  • National Institutes of Health (DGE-1256082, DK107979)
  • National Heart, Lung, and Blood Institute (T32HL007828, DGE-1256082)
  • National Human Genome Research Institute (DP1HG007811, R01HG006283, T32HG000035)
  • National Institute of Diabetes and Digestive and Kidney Diseases (U54DK107979)
  • Eunice Kennedy Shriver National Institute of Child Health and Human Development (DP2HD088158)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Chromatin Dynamics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Chromatin Remodeling and Cancer
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
Keywords
chromatin accessibility, myogenesis, Cicero, cis-regulation, MyoD, histone modifications, transcription factors, single-cell chromatin accessibility

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: Chromatin accessibility dynamics of myogenesis at single cell resolution. https://app.cochat.ai/papers/10.1101/155473
The paper
Pliner, H. A., Packer, J., McFaline-Figueroa, J. L., Cusanovich, D. A., Daza, R., Srivatsan, S., Qiu, X., Jackson, D., Minkina, A., Adey, A., Steemers, F. J., Shendure, J., & Trapnell, C. (2017). Chromatin accessibility dynamics of myogenesis at single cell resolution. bioRxiv (Cold Spring Harbor Laboratory). https://doi.org/10.1101/155473

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.