Alignment free multiple sequence analysis

CREATED: 200910031111

Given a set of sequences, $\mathcal{S}$, consider all subwords of length from 3 to 20, say.
Spread of a word: number of sequences it appears in
Range of a word: set of sequences it appears in
Consider word length vs spread, select relevant words, eg. unique spread
Each word selects a subset of sequences! Avoids dealing with numbers such as with distances
Local decoding: connect two words if they share a subword of length more than half