CoChat

MATCHER: manifold alignment reveals correspondence between single cell transcriptome and epigenome dynamics

Open access

Published version of a preprint

The DOI matches its Crossref record; 57 of the 58 references it cites were found in a registry or an index.

Abstract

Single cell experimental techniques reveal transcriptomic and epigenetic heterogeneity among cells, but how these are related is unclear. We present MATCHER, an approach for integrating multiple types of single cell measurements. MATCHER uses manifold alignment to infer single cell multi-omic profiles from transcriptomic and epigenetic measurements performed on different cells of the same type. Using scM&T-seq and sc-GEM data, we confirm that MATCHER accurately predicts true single cell correlations between DNA methylation and gene expression without using known cell correspondences. MATCHER also reveals new insights into the dynamic interplay between the transcriptome and epigenome in single embryonic stem cells and induced pluripotent stem cells.

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 · 10.1186/s13059-017-1269-0

Cites

57 of 57 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 MATCHER method overview. a We infer manifold representations of each dataset using a Gaussian process latent variable model (GPLVM). However, the resulting “pseudotime” values from different genomic data types are not directly comparable due to differences in orientation, scale, and “time warping.” Both the color of the curve (black to yellow) and cell morphology (blob to oblong) indicate position within this hypothetical process. b, c To account for these effects, pseudotime for each kind of data is modeled as a non-linear function (warping function) of master time using a Gaussian process. d MATCHER infers “master time” in which the steps of a biological process correspond to values uniformly distributed between 0 and 1 and are comparable among different data types. However, different datasets are measured from different physical cells and thus may sample different points in the biological process and even different numbers of cells. e Diagram showing how MATCHER’s generative model can infer corresponding cell measurements. The generated cell is drawn with transparency to indicate that this is an inferred rather than observed quantity. f Applying MATCHER to multiple types of data provides exactly corresponding measurements from observed cells and unobserved cells (indicated with transparency) generated by MATCHER
Fig. 1 MATCHER method overview. a We infer manifold representations of each dataset using a Gaussian process latent variable model (GPLVM). However, the resulting “pseudotime” values from different genomic data types are not directly comparable due to differences in orientation, scale, and “time warping.” Both the color of the curve (black to yellow) and cell morphology (blob to oblong) indicate position within this hypothetical process. b, c To account for these effects, pseudotime for each kind of data is modeled as a non-linear function (warping function) of master time using a Gaussian process. d MATCHER infers “master time” in which the steps of a biological process correspond to values uniformly distributed between 0 and 1 and are comparable among different data types. However, different datasets are measured from different physical cells and thus may sample different points in the biological process and even different numbers of cells. e Diagram showing how MATCHER’s generative model can infer corresponding cell measurements. The generated cell is drawn with transparency to indicate that this is an inferred rather than observed quantity. f Applying MATCHER to multiple types of data provides exactly corresponding measurements from observed cells and unobserved cells (indicated with transparency) generated by MATCHERCC BY · PubMed Central
Fig. 2 Single cell transcriptome and epigenome data show common modes of variation. a–d Single cell trajectories constructed by SLICER from RNA-seq, bisulfite sequencing, ATAC-seq, and H3K4me2 ChIP-seq of mESCs grown in serum. e–l Levels of important gene expression, DNA methylation, chromatin accessibility, and H3K4me2 markers across the trajectories. Note: We used SLICER for the analysis in this figure because it is a previously published method for constructing cell trajectories that allowed us to investigate the hypothesis that single cell transcriptome and epigenome measurements share common sources of variation. SLICER and MATCHER are completely separate methods; MATCHER does not rely on SLICER in any way; and SLICER could not be used to integrate multiple types of measurements as MATCHER does, because SLICER lacks the ability to generate unobserved cell measurements
Fig. 2 Single cell transcriptome and epigenome data show common modes of variation. a–d Single cell trajectories constructed by SLICER from RNA-seq, bisulfite sequencing, ATAC-seq, and H3K4me2 ChIP-seq of mESCs grown in serum. e–l Levels of important gene expression, DNA methylation, chromatin accessibility, and H3K4me2 markers across the trajectories. Note: We used SLICER for the analysis in this figure because it is a previously published method for constructing cell trajectories that allowed us to investigate the hypothesis that single cell transcriptome and epigenome measurements share common sources of variation. SLICER and MATCHER are completely separate methods; MATCHER does not rely on SLICER in any way; and SLICER could not be used to integrate multiple types of measurements as MATCHER does, because SLICER lacks the ability to generate unobserved cell measurementsCC BY · PubMed Central
Fig. 3 MATCHER accurately infers known correlations between DNA methylation and gene expression. a–c Heatmaps comparing true correlations between gene expression and DNA methylation of related regions (H3K27me3 peaks, LMRs, and P300 binding sites). The first column of each heatmap shows the true correlation based on known correspondence information, the second column shows the correlation inferred by MATCHER in the same dataset, and the third column is correlation inferred by MATCHER using a completely different single cell RNA-seq dataset from mESCs grown in serum. d, e Scatterplot representation of the results shown in (a–c). d Correlations computed using the Angermueller data. e Correlations computed from the Kolodziejczyk data. Each point represents the true and inferred correlation for a single gene-site pair; ideal results would lie along the y = x line. Note that the sign of the inferred correlation is correct for the vast majority of pairs
Fig. 3 MATCHER accurately infers known correlations between DNA methylation and gene expression. a–c Heatmaps comparing true correlations between gene expression and DNA methylation of related regions (H3K27me3 peaks, LMRs, and P300 binding sites). The first column of each heatmap shows the true correlation based on known correspondence information, the second column shows the correlation inferred by MATCHER in the same dataset, and the third column is correlation inferred by MATCHER using a completely different single cell RNA-seq dataset from mESCs grown in serum. d, e Scatterplot representation of the results shown in (a–c). d Correlations computed using the Angermueller data. e Correlations computed from the Kolodziejczyk data. Each point represents the true and inferred correlation for a single gene-site pair; ideal results would lie along the y = x line. Note that the sign of the inferred correlation is correct for the vast majority of pairsCC BY · PubMed Central
Fig. 4 Correlations among single cell gene expression, chromatin accessibility, and histone modifications. a Violin plot of correlations among chromatin accessibility and H3K4me2 of transcription factor binding sites for 186 transcription factors. Note that most correlations are strongly positive. b Correlation between chromatin accessibility and H3K4me2 data reveals that targets of pluripotency factors/NuRD complex and targets of Polycomb Group/Trithorax Group proteins are anticorrelated in single cells. c Correlation between gene expression signatures and chromatin accessibility signatures. d Correlation between gene expression signatures and H3K4me2 signatures. e Correlation between gene expression of DNA binding proteins and chromatin accessibility of their targets. f Inferred corresponding values of Sox2 gene expression and chromatin accessibility of SOX2 binding sites. Each point represents inferred correspondence from a single cell. The x-axis shows the value of the gene expression signature in that cell and the y-axis shows the value of the chromatin accessibility signature. The points are colored by inferred master time. g Inferred corresponding values of Yy1 gene expression and chromatin accessibility of YY1 binding sites
Fig. 4 Correlations among single cell gene expression, chromatin accessibility, and histone modifications. a Violin plot of correlations among chromatin accessibility and H3K4me2 of transcription factor binding sites for 186 transcription factors. Note that most correlations are strongly positive. b Correlation between chromatin accessibility and H3K4me2 data reveals that targets of pluripotency factors/NuRD complex and targets of Polycomb Group/Trithorax Group proteins are anticorrelated in single cells. c Correlation between gene expression signatures and chromatin accessibility signatures. d Correlation between gene expression signatures and H3K4me2 signatures. e Correlation between gene expression of DNA binding proteins and chromatin accessibility of their targets. f Inferred corresponding values of Sox2 gene expression and chromatin accessibility of SOX2 binding sites. Each point represents inferred correspondence from a single cell. The x-axis shows the value of the gene expression signature in that cell and the y-axis shows the value of the chromatin accessibility signature. The points are colored by inferred master time. g Inferred corresponding values of Yy1 gene expression and chromatin accessibility of YY1 binding sitesCC BY · PubMed Central
Fig. 5 Relationship between gene expression and DNA methylation in transition from ground state to primed pluripotency. a Scatterplot showing the relationship between master time inferred from gene expression and master time inferred from DNA methylation. Points are colored by the log10 expression of Rex1. The dotted line is the y = x line. Note that the gene expression and DNA methylation master time values are more correlated before master time = 0.3 than after. b, c Density plots showing the distribution of pseudotime inferred from b gene expression and c DNA methylation. The vertical dotted line indicates the 30th percentile of pseudotime (master time = 0.3). d Violin plot showing the distribution of Rex1 expression in cells before master time = 0.3 (“early”) and after master time = 0.3 (“late”). e Expression of Dnmt3b as a function of gene expression master time. The red line is a LOESS smoothing function indicating the overall expression trend. The black vertical line indicates master time = 0.3. f Expression of Tet1 as a function of gene expression master time. The red line is a LOESS smoothing function indicating the overall expression trend
Fig. 5 Relationship between gene expression and DNA methylation in transition from ground state to primed pluripotency. a Scatterplot showing the relationship between master time inferred from gene expression and master time inferred from DNA methylation. Points are colored by the log10 expression of Rex1. The dotted line is the y = x line. Note that the gene expression and DNA methylation master time values are more correlated before master time = 0.3 than after. b, c Density plots showing the distribution of pseudotime inferred from b gene expression and c DNA methylation. The vertical dotted line indicates the 30th percentile of pseudotime (master time = 0.3). d Violin plot showing the distribution of Rex1 expression in cells before master time = 0.3 (“early”) and after master time = 0.3 (“late”). e Expression of Dnmt3b as a function of gene expression master time. The red line is a LOESS smoothing function indicating the overall expression trend. The black vertical line indicates master time = 0.3. f Expression of Tet1 as a function of gene expression master time. The red line is a LOESS smoothing function indicating the overall expression trendCC BY · PubMed Central
Fig. 6 Analysis of gene expression and DNA methylation in human fibroblast cells undergoing reprogramming. a, b Density plots showing distribution of pseudotime inferred from a gene expression and b DNA methylation. The pseudotime values for individual cells are shown as a rug plot below the density plot; color indicates the time point. c Relationship between master time inferred from gene expression and master time inferred from DNA methylation. d Heatmap of ground truth correlation between expression of all genes measured in the sc-GEM experiment and DNA methylation level of all promoters measured. e Heatmap of correlation inferred by MATCHER from sc-GEM data. Note that MATCHER inferred these correlations without using the known correspondence among cells in any way. f Violin plot of the DNA methylation master time values for cells at each time point. Note that the distributions for untreated fibroblasts (BJ) and fibroblasts eight days after treatment (d8) are virtually identical. g Violin plot of the gene expression master time values for cells at each time point
Fig. 6 Analysis of gene expression and DNA methylation in human fibroblast cells undergoing reprogramming. a, b Density plots showing distribution of pseudotime inferred from a gene expression and b DNA methylation. The pseudotime values for individual cells are shown as a rug plot below the density plot; color indicates the time point. c Relationship between master time inferred from gene expression and master time inferred from DNA methylation. d Heatmap of ground truth correlation between expression of all genes measured in the sc-GEM experiment and DNA methylation level of all promoters measured. e Heatmap of correlation inferred by MATCHER from sc-GEM data. Note that MATCHER inferred these correlations without using the known correspondence among cells in any way. f Violin plot of the DNA methylation master time values for cells at each time point. Note that the distributions for untreated fibroblasts (BJ) and fibroblasts eight days after treatment (d8) are virtually identical. g Violin plot of the gene expression master time values for cells at each time pointCC BY · PubMed Central
Fig. 7 Incorporating known cell correspondence information to compute shared master time. a Scatterplot of shared master time inferred from both gene expression and DNA methylation (x-axis) and master time inferred using DNA methylation only (y-axis). b Scatterplot of shared master time inferred from both gene expression and DNA methylation (x-axis) and master time inferred using gene expression only (y-axis). c Plot showing “lagging cells” whose shared master time values overlap with the master time values of a previous time point. The x-values are jittered to mitigate overplotting. Colored horizontal lines indicate the maximum master time value for the corresponding time point. Lagging cells are indicated by “x” symbols. d Plot showing differences between lagging cells identified from shared master time and lagging cells identified from gene expression master time alone. The “x” symbols indicate lagging cells identified using shared master time. Arrows indicate two cells that are lagging based on gene expression master time alone but not shared master time. e Plot showing differences between lagging cells identified from shared master time and lagging cells identified from DNA methylation master time alone. The “x” symbols indicate lagging cells identified using shared master time
Fig. 7 Incorporating known cell correspondence information to compute shared master time. a Scatterplot of shared master time inferred from both gene expression and DNA methylation (x-axis) and master time inferred using DNA methylation only (y-axis). b Scatterplot of shared master time inferred from both gene expression and DNA methylation (x-axis) and master time inferred using gene expression only (y-axis). c Plot showing “lagging cells” whose shared master time values overlap with the master time values of a previous time point. The x-values are jittered to mitigate overplotting. Colored horizontal lines indicate the maximum master time value for the corresponding time point. Lagging cells are indicated by “x” symbols. d Plot showing differences between lagging cells identified from shared master time and lagging cells identified from gene expression master time alone. The “x” symbols indicate lagging cells identified using shared master time. Arrows indicate two cells that are lagging based on gene expression master time alone but not shared master time. e Plot showing differences between lagging cells identified from shared master time and lagging cells identified from DNA methylation master time alone. The “x” symbols indicate lagging cells identified using shared master timeCC 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 57 references

Cited by

See all 158

Metadata

Authors
Welch, J. D., Hartemink, A. J., & Prins, J. F.
Affiliations
  • University of North Carolina at Chapel Hill (US)
  • Duke University (US)
Corresponding author
Prins, Jan F.
Venue
Genome Biology
Published
2017-07-24
Volume
18
Issue
1
Pages
138-138
ISSN
  • 1465-6906 (ISSN-L)
  • 1474-760X
OpenAlex
W2950030509
Cited by
140
References
58
Citations by year
2026: 3 · 2025: 10 · 2024: 8 · 2023: 20 · 2022: 32 · 2021: 26 · 2020: 16 · 2019: 13 · 2018: 11 · 2017: 1
FWCI
4.58
Type
journal article
Language
en
Open access
Licence
Locations
  • Genome biology (publishedVersion, cc-by)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, other-oa)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Carolina Digital Repository (University of North Carolina at Chapel Hill) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
National Human Genome Research Institute (1F31HG008912-01, R01HG006272)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Epigenetics and DNA Methylation
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Chromatin Dynamics
Keywords
manifold alignment, single-cell multi-omics integration, DNA methylation, gene expression, embryonic stem cells, induced pluripotent stem cells, epigenetic heterogeneity
MeSH terms
Mouse Embryonic Stem Cells, Algorithms, Animals, Histones, Humans, Genome, Human, Sequence Analysis, RNA, DNA Methylation, Epigenesis, Genetic, Mice, Induced Pluripotent Stem Cells, Single-Cell Analysis, Transcriptome

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: MATCHER: manifold alignment reveals correspondence between single cell transcriptome and epigenome dynamics. https://app.cochat.ai/papers/10.1186/s13059-017-1269-0
The paper
Welch, J. D., Hartemink, A. J., & Prins, J. F. (2017). MATCHER: manifold alignment reveals correspondence between single cell transcriptome and epigenome dynamics. Genome Biology. https://doi.org/10.1186/s13059-017-1269-0

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.