Figure 3
From: Efficient pairwise RNA structure prediction and alignment using sequence alignment constraints
Constrained Sankoff Algorithm. Panel A: A cartoon depicting an example problem with four pins shown as dumbbells, labeled q z to qz+3, which connect the two sequences. The indices i, a, and j are positions in the first sequence x; k, b, and l are positions in the second sequence y. The corresponding segment edges L and R for each position in x are labeled. In this notation, j is the potential base pairing partner of i, l is the potential base pairing partner of k, and the subsequence i...j aligns with k...l. The indices a and b are the required for identifying bifurcation points. Panel B: The constrained structural alignment algorithm in pseudocode, where M is the length of sequence x, L (i) is the left edge of the segment containing the position i, and R (i) is the right edge of the segment containing i. The max in the range for l is required to handle the case where i and j share the same segment. The b range must be handled similarly. Panel C: The special case where the position under consideration is equivalent to a pin. In this case, we know the location of its alignment partner but must also consider the possibility of insertions in y which may occur before or after this pin.