Fig. 2

Some relevant design choices for pseudoalignment with skipping heuristics that jump and skip \(k\)-mers on a given read. After \(k\)-mer \(x_1\) is queried and found to map to a “black” unitig, an algorithm can jump to query the \(k\)-mer \(x_2\) on input read, where the number of \(k\)-mers skipped is given by the length of the black unitig. A In the ideal scenario, \(x_2\) maps to the black unitig sequence and \(k\)-mers \(x_1\) and \(x_2\) are found to bookend this unitig sequence as it appears on the read. B If \(x_2\) misses the index, an algorithm can back-off to an earlier \(k\)-mer on the read to find a \(k\)-mer bookending a shorter subsequence of the black unitig; or it may just query the next \(k\)-mer. C If \(x_2\) maps to a different “red” unitig, an algorithm has an alternative, aggressive, heuristic option to jump and find the next \(k\)-mer bookending the red unitig sequence