How to take RNA folding into account?

[Sjors]

I suspect (see #3) that RNA folding may play a role in the optimisation.

https://academic.oup.com/nar/article/48/22/12436/5961787

This may be of help: https://www.nature.com/articles/s41467-020-19371-y

[julianghionoiu]

I guess the question is what is better for expression, a tightly packed RNA molecule or a loose one?

There must be a mechanism that cleaves and recycles RNA so maybe a tightly packed molecule might avoid that.
At the same time, you don't want to get in the way of the transcription process.

On a separate note, maybe taking a measure such as the minimum free energy of the RNA molecule into account might yield some results.
Found this Python lib but have not yet tried it: https://www.tbi.univie.ac.at/RNA/ViennaRNA/doc/html/examples_python.html

[Sjors]

Perhaps Pfizer used a heuristic like "don't mess with the folding". Folded parts will have a matching sequence in reverse, e.g. ATGATG....CATCAT You can still swap those codons for something with more C and G, but you have to take the mirror codon into consideration.

So an algo would:

• find codon pairs (the hard part)
• optimize unpaired codons (see my PR)
• optimize paired codons pairwise