Quesition about Relax Mode

There’s no reason relax can’t be carried out on multimers, but it’s not intended for use on multi chain systems. The only real reason is that it doesn’t “know” there are multiple chains and will not attempt to maintain any particular relationship between them – as you describe it is ignorant of quaternary structure. If you used it and it worked then there’s no reason to doubt the result.

One other issue is that Rosetta works in internal coordinates, and depending on how the multiple chains are related to one another, single-chain protocols like relax can do terrible things when multiple chains are present. In internal coordinates, the later chains may be kinematically dependent on the first chain, so all changes made to it propagate down to later chains. This would mean small moves to the first chain would cause later chains to swing around in rigid-body space, which is bad. This can be fixed by having later chains all kinematically dependent directly on the root atom of the AtomTree (in other words, a tree structure rather than daisy-chained). This is the default setup for Rosetta to avoid the problem I described, so it usually doesn’t happen. Relax mode isn’t written to worry about it one way or another, so it definitely can cause problems under some circumstances.

I’m not aware of anything specialized for multiple chain relax. The relax code worked fine for you, and I think it will work fine almost all of the time – I was just trying to point out what the possible problems would be. Good luck!