[BI] Tools for chromatin-state discovery and annotation

A variety of strategic tools for chromatin state segmentation have been developed over time. Among them, ChromHMM remains the most widely used and recognized. However, many other tools have been introduced to address its limitations, often offering superior performance under specific conditions. Below is a comparative overview of these tools, highlighting their key features and reference citations.

🧠 Comparative Table of Epigenomic Segmentation Tools

Tool	Modeling Strategy	Key Features	Strengths	Limitations
ChromHMM	Multivariate HMM + EM	Learns chromatin states from binary histone mark data	Fast, easy to use, interpretable, widely adopted	Assumes same state model across samples, no cross-cell modeling
TreeHMM	Tree-structured HMM	Models lineage relationships among cell types	Captures developmental hierarchy, improves accuracy for related cells	Requires a known or assumed cell lineage tree
GATE	Graph-aware HMM	Integrates spatial proximity data (e.g., Hi-C)	Accounts for chromatin 3D structure	Depends on high-quality Hi-C or interaction data
diHMM	Hierarchical HMM	Models chromatin at both nucleosome and domain levels	Multi-scale annotation of genome	Computationally intensive, more complex training
CMINT	Bayesian mixture model	Jointly clusters cell types and learns chromatin states	Handles cell type heterogeneity	Model complexity, requires cluster number tuning
hiHMM	Hierarchical HMM	Models shared and cell-specific states across cell types	Captures both conserved and unique patterns	Assumes predefined hierarchical relationships
IDEAS	2D HMM (Bayesian, nonparametric)	Jointly models genome position × cell type dynamics explicitly	Cross-cell comparison, flexible state sharing, state number auto-inferred	Complex model, higher computational cost
Segway-GBR	Dynamic Bayesian network + Graph-based regularization	Integrates genome annotation with 3D genome architecture (Hi-C)	Models long-range interactions	Requires Hi-C and advanced setup
STAN	Bayesian changepoint detection	Focuses on detecting domain boundaries	Effective for identifying sharp transitions	Not designed for general-purpose segmentation
GenoSTAN	Poisson/Gaussian HMM	Models read counts from ChIP-seq directly	Good for quantitative signal modeling	Assumes statistical distributions; heavier computation
EpiCSeg	HMM + Count data	Uses actual read counts instead of binarization	More accurate modeling of weak/moderate signals	Slower performance, harder to interpret
Spectacle	Spectral learning-based HMM	Uses method-of-moments for fast parameter estimation	Faster than EM, reduces overfitting to null states	Spectral method requires careful initialization

📚 References for Each Tool

Tool name (citations; accessed July 1, 2025)

1. ChromHMM (n = 2702, 862)

   • Ernst J & Kellis M (2012) ChromHMM: automating chromatin-state discovery and characterization. Nature Methods

   • Ernst J & Kellis M (2017) Chromatin-state discovery and genome annotation with ChromHMM. Nature Protocols.

2. TreeHMM (n = 60)

   • Biesinger J et al. (2013) Discovering and mapping chromatin states using a tree hidden Markov model. BMC Bioinformatics.

3. GATE (n = 79)

   • Yu P et al. (2013) Spatiotemporal clustering of the epigenome reveals rules of dynamic gene regulation. Genome Res.

4. diHMM (n = 52)

   • Marco E et al. (2017) Multi-scale chromatin state annotation using a hierarchical hidden Markov model. Nat. Commun.

5. CMINT (n = 16)

   • Roy S & Sridharan R (2017) Chromatin module inference on cellular trajectories identifies key transition points and poised epigenetic states in diverse developmental processes. Genome Res.

6. hiHMM (n = 53)

   • Sohn K-A et al. (2015) hiHMM: Bayesian non-parametric joint inference of chromatin state maps. Bioinformatics.

7. IDEAS (n = 103)

   • Zhang Y et al. (2016) Jointly characterizing epigenetic dynamics across multiple human cell types. Nucleic Acids Res.

8. Segway-GBR (n = 89)

   • Libbrecht MW et al. (2015) Joint annotation of chromatin state and chromatin conformation reveals relationships among domain types and identifies domains of cell-type-specific expression. Genome Res.

9. STAN (n = 33)

   • Zacher B et al. (2014) Annotation of genomics data using bidirectional hidden Markov models unveils variations in Pol II transcription cycle. Mol. Syst. Biol.

10. GenoSTAN (n = 96)

    • Zacher B et al. (2017) Accurate promoter and enhancer identification in 127 ENCODE and roadmap epigenomics cell types and tissues by GenoSTAN. PLoS ONE.
11. EpiCSeg (n = 103)
    • Mammana A & Chung H-R (2015) Chromatin segmentation based on a probabilistic model for read counts explains a large portion of the epigenome. Genome Biol.
12. Spectacle (n = 55)
    • Song J & Chen KC (2015) Spectacle: fast chromatin state annotation using spectral learning. Genome Biol.