Uniprot has annotation for subcellular location of protein domains. This topological domain information of proteins is under the
In most cases the subcellular location is assigned by rigorous sequence analysis both for the experimentally verified and predicted subcellular locations. Sometimes more reliable information is available...
... when the experimental technique used allows the unambiguous assignment of the transmembrane boundary to a particular position (X-ray crystallography, etc.), the ‘Sequence analysis’ qualifier is not added in the topological domain annotation. In this case, the positions of the topological domains can be propagated ‘By similarity’.
What evidence is available that constitutes "unambiguous assignment" so that the
by similarity qualifier can be used?