Is there any protocol for this as mentioned in the title?
My X-ray data do not have phase included and my NMR data do not have NOESY data.
CS-Rosetta is basically ab initio, biased by using chemical shift to generate fragments. I guess you could run CS-Rosetta to fold your protein, then see if you can use those as MR phasing candidates. I don't think that Rosetta (or anyone) can do much with phaseless crystal data by itself...
There's work underway to better interface Phenix into Rosetta so that Rosetta can internally pass structure factors and a proposed model to Phenix to generate a density map. This way Rosetta can do the real space refinement, and interfaces with Phenix to automatically handle the re-estimation of phases and map rebuilding. - I don't believe this is published yet, though, and you'd need an initial molecular replacement model to get things started.
As Steven indicates, a good approach is to use CS-Rosetta generated models as MR starting points. (You've seen the recent Khatib et al. paper, right? We're suggesting something similar, but with CS-Rosetta generated structures rather than Foldit results.) Once you've gotten the initial phased electron density map, you may be able to get better results by manually iterating Rosetta refinement and map rebuilding (paper).