I have a protein whose sequence is fully known and whose structure is partial known. Most sequences of the protein are folded in the luminal ER domain and this part is resolved by experiments. However, transmembrane and cytosolic parts of the same protein are not resolved. As only the luminal part of the protein is resolved, I would like to predict the structure for transmembrane and cytosolic domains. Is there a way to predict only these parts while I keep the experimentally-resolved luminal domain?