That particular error is signaling that when trying to build a particular protein, you’re trying to replace an atom in the FoldTree/AtomTree, but the old atom you’re trying to replace is not present. This can happen when you’re deleting or inserting residues.
Is this error happening in the threading step, or the silent file extraction step? Is it specific to the mpi version, or do you see it with a single-processor run? Are there particular proteins/alignments which are giving you problems? And what’s the exact commandline that’s giving you issues?
Is there anything special about the sequences you’re threading and the structures you’re threading them onto? Are there post-translation modifications? Are any of the input structures multi-chain? Do they have any residues which Rosetta isn’t recognizing appropriately? (There may be a message printed to the logging output earlier on.) How complex is your threading? Is it relatively simple, or are there a lot of small gaps or small segments of protein (especially one residue segments) interspersed with each other?
If you’re able to put together an example of a short (preferably single-processor) run which shows the problem, and post the commandline and input files here, I may be able to take a look at things and get a better sense of what’s wrong.