Exactly why you want to study it with MARTINI? MARTINI restrain secondary structure dynamics to that of the original template. For siRNA where conformational dynamics is essential it is a serious burden. Other than that, there is no RNA parameters developed yet. DNA parameters were released by S. Marrink group recently

DOI: 10.1021/acs.jctc.5b00286

also targeting large duplexes, not small oligonucleotides. These you can do far better with all-atom MD. If you positively need it then, there was recent extension for si-RNA which you would have to build by hands:

Systematic coarse-grained modeling of complexation between small interfering RNA and polycations

http://dx.doi.org/10.1063/1.4937384

thank you very much for your thoughts Dr. Sergei. i'm in need of simulation of siRNA-Protein complex and have to choose CG due to computational cost (on a single computer). anyway, im going to try all-atom MD first.