Trend AnalysisBiology & Life Sciences
Single-Cell Multi-Omics: Integrating Genome, Epigenome, and Transcriptome in One Cell
Single-cell RNA sequencing revolutionised our understanding of cellular heterogeneity, but gene expression is only one layer of cellular identity. Chromatin accessibility (ATAC-seq), DNA methylation, ...
By Sean K.S. Shin
This blog summarizes research trends based on published paper abstracts. Specific numbers or findings may contain inaccuracies. For scholarly rigor, always consult the original papers cited in each post.
The Question
Single-cell RNA sequencing revolutionised our understanding of cellular heterogeneity, but gene expression is only one layer of cellular identity. Chromatin accessibility (ATAC-seq), DNA methylation, histone modifications, and protein abundance all contribute to cell state. Single-cell multi-omics technologies now measure two or more of these layers simultaneously in the same individual cell, enabling direct interrogation of regulatory relationships: which chromatin changes drive which gene expression changes in which cell types? But the computational challenge is formidable โ how do you integrate sparse, noisy measurements across modalities to extract biological insight?
Landscape
T. Wang et al. (2024) in Nature Cell Biology, applied single-cell multi-omics (transcriptome + chromatin accessibility + DNA methylation) to human preimplantation embryos, identifying cytoskeletal defects during embryonic arrest. This study demonstrated the power of multi-omics at the single-cell level: transcriptomic changes alone would not have revealed the epigenetic dysregulation underlying arrested development.
C. Huang et al. (2024) created scCancerExplorer โ a pan-cancer database for interactively exploring single-cell multi-omics data across cancer types. By integrating genomic, epigenomic, and transcriptomic alterations at single-cell resolution, the database enables researchers to identify cancer-type-specific regulatory programmes and potential therapeutic targets.
Computationally, Mao et al. (2025) developed HALO โ a hierarchical causal modelling framework that goes beyond correlation to infer causal relationships between chromatin accessibility and gene expression. Their key insight: open chromatin does not always predict active transcription; the relationship is modality-specific and context-dependent.
H. Li et al. (2025), published in Nature Methods, introduced scMultiSim โ a simulation tool for benchmarking single-cell multi-omics integration methods. By generating realistic synthetic data with known ground truth, scMultiSim enables fair comparison of the rapidly growing number of computational integration methods.
Key Claims & Evidence
<
| Claim | Evidence | Verdict |
|---|
| Multi-omics reveals regulatory mechanisms invisible to single-modality studies | Epigenetic dysregulation in embryonic arrest detected only by multi-omics (T. Wang et al. 2024) | Strongly supported |
| Pan-cancer multi-omics databases enable cross-cancer comparisons | scCancerExplorer integrates genomic + epigenomic + transcriptomic data (C. Huang et al. 2024) | Demonstrated; data quality varies across studies |
| Open chromatin does not always predict active transcription | Causal modelling reveals modality-specific, context-dependent relationships (Mao et al. 2025) | Important nuance; challenges simplistic interpretations |
| Simulation tools are needed for method benchmarking | scMultiSim generates ground-truth multi-omics data (H. Li et al. 2025) | Confirmed; essential for methodological rigour |
Open Questions
Scalability: Current multi-omics experiments measure thousands of cells. Can throughput reach millions while maintaining data quality across all modalities?
Protein-level integration: Most single-cell multi-omics pairs transcriptome + epigenome. Can proteomic measurements (CITE-seq, single-cell mass spec) be added as a third simultaneous layer?
Temporal resolution: Multi-omics captures snapshots. Can time-series multi-omics (perturbation followed by sequential sampling) reveal causal chains in development and disease?
Clinical application: Can single-cell multi-omics profiling of patient tumours guide treatment decisions, or is the cost and complexity prohibitive for clinical use?Referenced Papers
- [1] Huang, C. et al. (2024). scCancerExplorer: pan-cancer single-cell multi-omics database. bioRxiv. DOI: 10.1101/2024.06.24.600356
- [2] Wang, T. et al. (2024). Single-cell multi-omics of human preimplantation embryos. Nature Cell Biology. DOI: 10.1038/s41556-023-01328-0
- [3] Shi, Z. et al. (2025). MomicPred: Cell Cycle Prediction from Single-Cell Multi-Omics. IEEE J. Biomedical and Health Informatics. DOI: 10.1109/JBHI.2025.3595904
- [4] Li, H. et al. (2025). scMultiSim: simulation of single-cell multi-omics and spatial data. Nature Methods. DOI: 10.1038/s41592-025-02651-0
- [5] Mao, H. et al. (2025). HALO: hierarchical causal modeling for single-cell multi-omics. Nature Communications. DOI: 10.1038/s41467-025-63921-1
๋ฉด์ฑ
์กฐํญ: ์ด ๊ฒ์๋ฌผ์ ์ ๋ณด ์ ๊ณต ๋ชฉ์ ์ ์ฐ๊ตฌ ๋ํฅ ๊ฐ์์ด๋ค. ํ์ ์ฐ๊ตฌ์์ ์ธ์ฉํ๊ธฐ ์ ์ ๊ตฌ์ฒด์ ์ธ ์ฐ๊ตฌ ๊ฒฐ๊ณผ, ํต๊ณ ๋ฐ ์ฃผ์ฅ์ ์๋ณธ ๋
ผ๋ฌธ์ ํตํด ํ์ธํด์ผ ํ๋ค.
๋จ์ผ์ธํฌ ๋ฉํฐ-์ค๋ฏน์ค: ํ๋์ ์ธํฌ์์ ๊ฒ๋, ์ํผ๊ฒ๋, ์ ์ฌ์ฒด์ ํตํฉ
๋ถ์ผ: ์๋ฌผํ | ๋ฐฉ๋ฒ๋ก : ๊ณ์ฐ-์คํ์
์ ์: Sean K.S. Shin | ๋ ์ง: 2026-03-17
์ฐ๊ตฌ ์ง๋ฌธ
๋จ์ผ์ธํฌ RNA ์ํ์ฑ์ ์ธํฌ ์ด์ง์ฑ์ ๋ํ ์ดํด๋ฅผ ํ์ ์ ์ผ๋ก ๋ณํ์์ผฐ์ผ๋, ์ ์ ์ ๋ฐํ์ ์ธํฌ ์ ์ฒด์ฑ์ ํ ์ธต์์ ๋ถ๊ณผํ๋ค. ํฌ๋ก๋งํด ์ ๊ทผ์ฑ(ATAC-seq), DNA ๋ฉํธํ, ํ์คํค ๋ณํ, ๋จ๋ฐฑ์ง ํจ๋ ๋ชจ๋ ์ธํฌ ์ํ์ ๊ธฐ์ฌํ๋ค. ๋จ์ผ์ธํฌ ๋ฉํฐ-์ค๋ฏน์ค ๊ธฐ์ ์ ์ด์ ๋์ผํ ๊ฐ๋ณ ์ธํฌ์์ ์ด๋ฌํ ์ธต์ ์ค ๋ ๊ฐ์ง ์ด์์ ๋์์ ์ธก์ ํจ์ผ๋ก์จ, ์กฐ์ ๊ด๊ณ๋ฅผ ์ง์ ์ ์ผ๋ก ๊ท๋ช
ํ ์ ์๊ฒ ํ๋ค: ์ด๋ค ํฌ๋ก๋งํด ๋ณํ๊ฐ ์ด๋ค ์ธํฌ ์ ํ์์ ์ด๋ค ์ ์ ์ ๋ฐํ ๋ณํ๋ฅผ ์ ๋ํ๋๊ฐ? ๊ทธ๋ฌ๋ ๊ณ์ฐ์ ๊ณผ์ ๋ ๋ง๋ง์น ์๋ค โ ์๋ฌผํ์ ํต์ฐฐ์ ๋์ถํ๊ธฐ ์ํด ์ฌ๋ฌ ๋ชจ๋ฌ๋ฆฌํฐ์ ๊ฑธ์น ํฌ๋ฐํ๊ณ ์ก์์ด ๋ง์ ์ธก์ ๊ฐ๋ค์ ์ด๋ป๊ฒ ํตํฉํ ๊ฒ์ธ๊ฐ?
์ฐ๊ตฌ ๋ํฅ
T. Wang ๋ฑ(2024)์ Nature Cell Biology์์ ๋จ์ผ์ธํฌ ๋ฉํฐ-์ค๋ฏน์ค(์ ์ฌ์ฒด + ํฌ๋ก๋งํด ์ ๊ทผ์ฑ + DNA ๋ฉํธํ)๋ฅผ ์ธ๊ฐ ์ฐฉ์ ์ ๋ฐฐ์์ ์ ์ฉํ์ฌ, ๋ฐฐ์ ๋ฐ๋ฌ ์ ์ง ์ค ์ธํฌ๊ณจ๊ฒฉ ๊ฒฐํจ์ ๊ท๋ช
ํ์๋ค. ์ด ์ฐ๊ตฌ๋ ๋จ์ผ์ธํฌ ์์ค์์ ๋ฉํฐ-์ค๋ฏน์ค์ ๊ฐ์ ์ ์
์ฆํ์๋ค: ์ ์ฌ์ฒด ๋ณํ๋ง์ผ๋ก๋ ๋ฐ๋ฌ ์ ์ง์ ๊ทผ๊ฐ์ด ๋๋ ํ์ฑ์ ์ ํ์ ์กฐ์ ์ด์์ ๋ฐํ๋ผ ์ ์์์ ๊ฒ์ด๋ค.
C. Huang ๋ฑ(2024)์ ์ ์ ํ์ ๊ฑธ์ณ ๋จ์ผ์ธํฌ ๋ฉํฐ-์ค๋ฏน์ค ๋ฐ์ดํฐ๋ฅผ ๋ํํ์ผ๋ก ํ์ํ๊ธฐ ์ํ ๋ฒ์ ๋ฐ์ดํฐ๋ฒ ์ด์ค์ธ scCancerExplorer๋ฅผ ๊ตฌ์ถํ์๋ค. ๋จ์ผ์ธํฌ ํด์๋์์ ๊ฒ๋, ์ํผ๊ฒ๋, ์ ์ฌ์ฒด ๋ณ์ด๋ฅผ ํตํฉํจ์ผ๋ก์จ, ์ด ๋ฐ์ดํฐ๋ฒ ์ด์ค๋ ์ฐ๊ตฌ์๋ค์ด ์ ์ ํ ํน์ด์ ์กฐ์ ํ๋ก๊ทธ๋จ๊ณผ ์ ์ฌ์ ์น๋ฃ ํ์ ์ ๊ท๋ช
ํ ์ ์๊ฒ ํ๋ค.
๊ณ์ฐ์ ์ธก๋ฉด์์, Mao ๋ฑ(2025)์ ํฌ๋ก๋งํด ์ ๊ทผ์ฑ๊ณผ ์ ์ ์ ๋ฐํ ๊ฐ์ ์ธ๊ณผ ๊ด๊ณ๋ฅผ ์๊ด๊ด๊ณ๋ฅผ ๋์ด ์ถ๋ก ํ๋ ๊ณ์ธต์ ์ธ๊ณผ ๋ชจ๋ธ๋ง ํ๋ ์์ํฌ์ธ HALO๋ฅผ ๊ฐ๋ฐํ์๋ค. ๊ทธ๋ค์ ํต์ฌ ํต์ฐฐ์ ๋ค์๊ณผ ๊ฐ๋ค: ์ด๋ฆฐ ํฌ๋ก๋งํด์ด ํญ์ ํ์ฑ ์ ์ฌ๋ฅผ ์์ธกํ์ง๋ ์์ผ๋ฉฐ, ๊ทธ ๊ด๊ณ๋ ๋ชจ๋ฌ๋ฆฌํฐ ํน์ด์ ์ด๊ณ ๋งฅ๋ฝ ์์กด์ ์ด๋ค.
H. Li ๋ฑ(2025)์ Nature Methods์ ๋จ์ผ์ธํฌ ๋ฉํฐ-์ค๋ฏน์ค ํตํฉ ๋ฐฉ๋ฒ์ ๋ฒค์น๋งํน์ ์ํ ์๋ฎฌ๋ ์ด์
๋๊ตฌ์ธ scMultiSim์ ๋ฐํํ์๋ค. ์๋ ค์ง ์ค์ธก ์๋ฃ(ground truth)๋ฅผ ๊ฐ๋ ํ์ค์ ์ธ ํฉ์ฑ ๋ฐ์ดํฐ๋ฅผ ์์ฑํจ์ผ๋ก์จ, scMultiSim์ ๋น ๋ฅด๊ฒ ์ฆ๊ฐํ๋ ๊ณ์ฐ์ ํตํฉ ๋ฐฉ๋ฒ๋ค์ ๊ณต์ ํ ๋น๊ต๋ฅผ ๊ฐ๋ฅํ๊ฒ ํ๋ค.
์ฃผ์ ์ฃผ์ฅ ๋ฐ ๊ทผ๊ฑฐ
<
| ์ฃผ์ฅ | ๊ทผ๊ฑฐ | ํ๊ฐ |
|---|
| ๋ฉํฐ-์ค๋ฏน์ค๋ ๋จ์ผ ๋ชจ๋ฌ๋ฆฌํฐ ์ฐ๊ตฌ์์๋ ๋ณด์ด์ง ์๋ ์กฐ์ ๋ฉ์ปค๋์ฆ์ ๋๋ฌ๋ธ๋ค | ๋ฐฐ์ ๋ฐ๋ฌ ์ ์ง์์์ ํ์ฑ์ ์ ํ์ ์กฐ์ ์ด์์ด ๋ฉํฐ-์ค๋ฏน์ค์ ์ํด์๋ง ๊ฒ์ถ๋จ (T. Wang ๋ฑ 2024) | ๊ฐ๋ ฅํ ์ง์ง๋จ |
| ๋ฒ์ ๋ฉํฐ-์ค๋ฏน์ค ๋ฐ์ดํฐ๋ฒ ์ด์ค๋ ์ ๊ฐ ๋น๊ต๋ฅผ ๊ฐ๋ฅํ๊ฒ ํ๋ค | scCancerExplorer๊ฐ ๊ฒ๋ + ์ํผ๊ฒ๋ + ์ ์ฌ์ฒด ๋ฐ์ดํฐ๋ฅผ ํตํฉํจ (C. Huang ๋ฑ 2024) | ์
์ฆ๋จ; ๋ฐ์ดํฐ ํ์ง์ ์ฐ๊ตฌ๋ง๋ค ๋ค๋ฆ |
| ์ด๋ฆฐ ํฌ๋ก๋งํด์ด ํญ์ ํ์ฑ ์ ์ฌ๋ฅผ ์์ธกํ์ง๋ ์๋๋ค | ์ธ๊ณผ ๋ชจ๋ธ๋ง์ด ๋ชจ๋ฌ๋ฆฌํฐ ํน์ด์ , ๋งฅ๋ฝ ์์กด์ ๊ด๊ณ๋ฅผ ๋๋ฌ๋ (Mao ๋ฑ 2025) | ์ค์ํ ์ธ๋ถ ์ฌํญ; ๋จ์ํ ํด์์ ์ด์๋ฅผ ์ ๊ธฐํจ |
| ๋ฐฉ๋ฒ ๋ฒค์น๋งํน์ ์ํ ์๋ฎฌ๋ ์ด์
๋๊ตฌ๊ฐ ํ์ํ๋ค | scMultiSim์ด ์ค์ธก ์๋ฃ๋ฅผ ํฌํจํ ๋ฉํฐ-์ค๋ฏน์ค ๋ฐ์ดํฐ๋ฅผ ์์ฑํจ (H. Li ๋ฑ 2025) | ํ์ธ๋จ; ๋ฐฉ๋ฒ๋ก ์ ์๋ฐ์ฑ์ ํ์์ ์ |
๋ฏธํด๊ฒฐ ์ง๋ฌธ
ํ์ฅ์ฑ: ํ์ฌ์ ๋ฉํฐ-์ค๋ฏน์ค ์คํ์ ์์ฒ ๊ฐ์ ์ธํฌ๋ฅผ ์ธก์ ํ๋ค. ๋ชจ๋ ๋ชจ๋ฌ๋ฆฌํฐ์ ๊ฑธ์ณ ๋ฐ์ดํฐ ํ์ง์ ์ ์งํ๋ฉด์ ์ฒ๋ฆฌ๋์ ์๋ฐฑ๋ง ๊ฐ๊น์ง ๋๋ฆด ์ ์๋๊ฐ?
๋จ๋ฐฑ์ง ์์ค์ ํตํฉ: ๋๋ถ๋ถ์ ๋จ์ผ์ธํฌ ๋ฉํฐ-์ค๋ฏน์ค๋ ์ ์ฌ์ฒด + ์ํผ๊ฒ๋์ ๊ฒฐํฉํ๋ค. ๋จ๋ฐฑ์ฒด ์ธก์ (CITE-seq, ๋จ์ผ์ธํฌ ์ง๋ ๋ถ์)์ ์ธ ๋ฒ์งธ ๋์ ์ธต์๋ก ์ถ๊ฐํ ์ ์๋๊ฐ?
์๊ฐ์ ํด์๋: ๋ค์ค ์ค๋ฏน์ค(multi-omics)๋ ์ค๋
์ท์ ํฌ์ฐฉํ๋ค. ์๊ณ์ด ๋ค์ค ์ค๋ฏน์ค(์ญ๋ ํ ์์ฐจ์ ์ํ๋ง)๊ฐ ๋ฐ๋ฌ ๋ฐ ์ง๋ณ์์ ์ธ๊ณผ์ ์ฐ์๋ฅผ ๋ฐํ๋ผ ์ ์๋๊ฐ?
์์ ์ ์ฉ: ํ์ ์ข
์์ ๋จ์ผ์ธํฌ ๋ค์ค ์ค๋ฏน์ค ํ๋กํ์ผ๋ง์ด ์น๋ฃ ๊ฒฐ์ ์ ์๋ดํ ์ ์๋๊ฐ, ์๋๋ฉด ๋น์ฉ๊ณผ ๋ณต์ก์ฑ์ด ์์์ ์ฌ์ฉ์ ์ง๋์น ์ฅ๋ฒฝ์ด ๋๋๊ฐ?References (5)
Huang, C., Liu, Z., Guo, Y., Wang, W., Yuan, Z., Guan, Y., et al. (2024). scCancerExplorer: a comprehensive database for interactively exploring single-cell multi-omics data of human pan-cancer.
Wang, T., Peng, J., Fan, J., Tang, N., Hua, R., Zhou, X., et al. (2024). Single-cell multi-omics profiling of human preimplantation embryos identifies cytoskeletal defects during embryonic arrest. Nature Cell Biology, 26(2), 263-277.
Shi, Z., Cong, L., & Wu, H. (2026). MomicPred: A Cell Cycle Prediction Framework Based on Dual-Branch Multi-Modal Feature Fusion for Single-Cell Multi-Omics Data. IEEE Journal of Biomedical and Health Informatics, 30(3), 2242-2251.
Li, H., Zhang, Z., Squires, M., Chen, X., & Zhang, X. (2025). scMultiSim: simulation of single-cell multi-omics and spatial data guided by gene regulatory networks and cellโcell interactions. Nature Methods, 22(5), 982-993.
Mao, H., Jia, M., Di, M., Valenzi, E., Cai, X. T., Lafyatis, R., et al. (2025). HALO: hierarchical causal modeling for single cell multi-omics data. Nature Communications, 16(1).