Fig. 4.

Lossless filtration illustrated by an example sequence of length 20 with τ = 3. The last τ nucleotides of the query are removed and the rest is divided into four series of contiguous segments. Each series is queried against a different index numbered I–IV. For instance, the only segment queried against index I is GTTG, whereas those queried against index II are GCAA, CAAT and AATA. If any of the segments is found in the appropriate index, the trie search is performed, otherwise it is omitted as there can be no τ-match. Regardless of the result, segments labeled I–IV are then added to the corresponding respective index (i.e. only one segment is added to each index)