R’HOM is a set of programs to segment a DNA sequence into a finite number of homogeneous regions.

The models used are Hidden Markov Models, HMM (M1-Mk). Segment arrangement is represented by an unobservable first order Markov chain, the hidden state chain (M1). The homogeneous regions are described by different models, more precisely by k-order Markov chains (Mk). Each model has characteristic statistical properties and take into account the local composition in oligonucleotides of length k+1 of the DNA sequence.

Several iterative algorithms are proposed to, on one hand, reconstruct the hidden state chain, that is to identify and locate the homogeneous regions of a given sequence (or a set of sequences), and on the other hand, to estimate the model parameters (state and observation transitions), in order to characterize these regions. When R’HOM is used on a set of sequences, the sequences are assumed to be independently generated using an unique set of parameters for all sequences (leading to a global estimation).