I have a set of RNA sequences, and I have been analyzing slices (i.e. target frames) of these sequences in order to determine whether they conform to certain structural criteria. One of these criteria requires that both the target frame of the RNA sequence and its flanking regions have as few secondary structures as possible, i.e. that the RNA nucleotides flanking the slice are as linear as possible.
Using Python and a number of standalone executables, I have been calculating the minimum free energy and optimal structural conformations of the RNA sequences and their flanking regions (see here for more details: http://math.mit.edu/classes/18.417/Slides/rna-prediction-zuker.pdf), but I realized that there is a slight hiccup in my current approach. Namely, I determine the secondary structures of the flanking regions as though they were sequences isolated from my target frame of interest. Now, given that the whole is more than just the sum its parts, I do not know if the secondary structures that are determined when the flanking region is examined on its own versus when the flanking region is examined including either a part or the whole target frame are the same.
Right now my code evaluates 20 nucleotides up and downstream of the target frame as the flanking regions, but should I instead, for example, evaluate 20 nucleotides up and downstream plus some nucleotides of my target frame? Right now, my target frame is 120 nucleotides long, so it might be worthwhile to consider 10 or 20 nucleotides of my target frame that are attached to the flanking regions. But I'm not sure how many nucleotides from my target length should be considered in order for the number of secondary structures to be consistent in the flanking regions. Or is this approach somehow faulty?