Transat—A Method for Detecting the Conserved Helices of Functional RNA Structures, Including Transient, Pseudo-Knotted and Alternative Structures
Transat first predicts the helices for all individual sequences in the fixed input alignment and then maps all of them to the alignment remembering the base-pairing sequence positions. In the example above, there are two helices, one derives from sequence 1 (see top figure), the other one from sequence 2. Mapping these two helices from their respective sequence to the entire alignment results in the two potential conserved helices shown above (see the arcs linking the respective alignment columns). Both conserved helices are then evaluated by Transat in terms of log-likelihood value and p-value estimation. The log-likelihood value is calculated based on the base-paired alignment columns in that helix and all sequences in the alignment, see the text and Figure 2 for details. All helices predicted by Transat for a given input alignment can then be ranked according to their p-value. For the two helices in the example above, the helix that fits the sequences in the given alignment better will have the higher log-likelihood value and lower p-value. As Transat is not capable of modifying the fixed input alignment, this mapping strategy minimized the impact of alignment errors.