Figure 10

Visualized CRF of a known polyadenylation signal motif. A StickWRLD view of the genomic sequence motif that governs “signal” based polyadenylation. Colors and categories are as previously shown. The motif representation starts in the back of the cylinder and proceeds counterclockwise. It is relatively easily modeled as 4“don’t care” positions with no significant base preference, followed by 2 A bases (red balls), a single T base (blue ball), and then three more A bases. There are a few very small residuals that attain statistical significance, but the marginal distributions dominate the motif, allowing it to be found using both PSSM and HMM methods.