When are the non-canonical base pairs, such as A:G, G:G and A:A going to be modeled in Eterna?
Currently, some of these are implicitly modeled when they show up inside internal loops, e.g.,:
C=G
G=C
G * A
A * G
C=G
G=C
The EteRNA folding rules, which are based on theViennaRNA folding engine, are calibrated on measurements by the Turner lab and others that include some of these sorts of internal bulges or loops. So even though G*A shows up in eterna (or other programs) as not base-paired, they get different energy bonuses then other internal loops.
The problem is how to more generally calculate/predict energies of these kinds of base pairs in any contexts, e.g. what if you had:
G=C
G A
A A
A G
A A
G G
C=G
which has not been measured in any context. Currently there’s only an approximate guess in our folding rules.
To be most useful, we’d want accuracies of better than 0.5 kcal/mol. This turns out to be a deep and open problem in computational biophysics, and its made especially intricate by the possibility of several alternative ways that G can pair with A, or A with A. Indeed at least 2 people in my lab are working on this… but I think the answer is at least 1-2 years away, if not further. We do have ways that involve 3D modeling to get approximate energies (with errors of about 1-2 kcal/mol), and are thinking of how to incorporate in future versions of eterRNA.