Figure 6:

A qualitative survey of repeat regions labeled by one tool but not others. (A) TRF–labeled repeats not found by ULTRA are typically made up 2–3 copies of a modest–length subunit. We have broken the examples into multiple rows to highlight the repeat (implied insertions are selected by hand). (B) TANTAN-labeled repeats not found by ULTRA are usually of the form shown here: frequently short, often composed of degenerate mono- or di-nucleotide repeats. Because of the simplicity of these repeats, they are not broken out across multiple lines. (C) ULTRA–labeled repeats not found by the other tools often match this form: a fairly long period with repeats containing non–trivial insertion or deletion structure. As with TRF repeats, the examples are broken across multiple lines to show the repetitive structure. In the case of the ULTRA “alignment” of repeats, the gap structure matches that used by the model to identify the repeat (i.e. it allows only consecutive insertions or deletions, and has not been optimized).