On another thread, it was said that EteRNA sequence designer starts with a sequence of 60% GC pairs. There might be other things that could help to generate a better initial sequence. Here was the idea that I was thinking about at the moment:
- Set closing pairs to GC.
- Set non-closing pairs to AU, placing random GC pairs to bring %GC up to 60 if necessary.
- Set terminal mismatches in hairpins to GA, with G placed 5’ of A.
- Set 1-1 loops to GG mismatches
- Set 2-1 loops to GG/A, GA/G,or AA/A with equal probability
- Set terminal mismatches in all other internal loops to GA.
- Set any other unpaired bases to A
Next, use Aldo’s mismatch strategy to look for regions that can form consecutive unwanted base pairs. If the number of unwanted pairs is larger than a certain cutoff, flip a random base pair in the region. Reevaluate the new sequence and flip another base pair if necessary. Repeat this process several times, and select the best sequence generated.
If anyone else has other ideas, please post them too…