Figure 8

Transposon data results. The table in (a) shows the average, over 16 runs from different starting profiles, of the number of transitive hits and the total number of matches and transitions in those transitive alignments. All three new algorithm variants show significant improvements over Baum-Welch, and the CBW algorithm exhibits the greatest improvement. The table in (b) shows the RepeatMasker results, results using HMMer-2 in conjunction with ClustalW and MUSCLE, and results using SAM with unaligned inputs and with aligned inputs.