Table 1.

Major differences between ChromHMM and Segway as applied to the ENCODE data

	ChromHMM	Segway
Modeling framework	Hidden Markov model	Dynamic Bayesian network
Genomic resolution	200 bp	1 bp
Data resolution	Boolean	Real value
Handling missing data	Interpolation	Marginalization
Emission modeling	Bernoulli distribution	Gaussian distribution
Length modeling	Geometric distribution	Geometric plus hard and soft constraints
Training set	Entire genome	ENCODE regions (1%)
Decoding algorithm	Posterior decoding	Viterbi
Learning across six cell types	Single model for all cell types	One model per cell type