Reiterating what I wrote in that thread - for an R-loop, the best option (at least until we release the next iteration of the nucleic acid FF ;)) would be to use DE Shaw’s RNA/DNA FFs due to the significant impact of the single-stranded part, and they certainly improve ssDNA/ssRNA description over alternatives. However, for that you have to ask them for the files directly, and they’re not in the most convenient format, so some knowledge of Gromacs technical details is necessary.
Meanwhile, in the user-contributed FFs, there is a ready-made combination of OL3 (for RNA) and OL15 (for DNA) that should work right away. You can also combine OL3 with parmbsc1 by preparing them separately, adding the parameters explicitly with gromologist, and merging the two molecules into one topology. These options should be much easier to implement but might yield a less flexible ssDNA region than desired; look at this paper for a discussion of how much less.