INVERTED-REPEATS-AWAREFINITE-CONTEXTMODELSFORDNACODING

Finite-context models have been used for DNA sequence compression as secondary, fall back mechanisms, the generalized opinion being that models with order larger than two or three are inappropriate. In this paper we show that finite-context models can also be used as the main encoding method, and that they are effective for model orders at least as higher as thirteen. Moreover, we propose a new model updating scheme that takes into account inverted repeats, a common characteristic in DNA sequences.