Transat—A Method for Detecting the Conserved Helices of Functional RNA Structures, Including Transient, Pseudo-Knotted and Alternative Structures
Transat takes as input a multiple sequence alignment and an evolutionary tree (left figure, top). It first predicts helices for all individual sequences in the alignment and then projects them back onto the multiple sequence alignment. It then calculates the log-likelihood value for each helix and estimates a p-value. The p-value estimation is explained in the right figure. In the first step, the original Transat input alignment is realigned based on primary sequence conservation only. In the second step, the columns of the resulting alignment are permuted multiple times, resulting in 500 shuffled versions of the original alignment. For each shuffled alignment, conserved helices are detected and their log-likelihood values calculated as for the original alignment. In the final step, the log-likelihood values of all helices in the shuffled alignments are entered into a histogram which is then used to derive p-values for the helices of the original alignment.