I am using homology modelling to assign the 3D structure of Torpedo acetylcholine receptor (Unwin 2004, 2bg9 in RCSB) to human muscle nAChR. The problem is, both Torpedo receptor, and human receptor have subunit structure A-B-D-A-E: so a same subunit (A) occurs in the model twice. This seems to cause problems for homology modelling software, as it tends to align the second 'A' sequence over the first 'A' sequence, creating a hole in the pentamer, and not filling it up. Any ideas on how to make sure the modelling software acknowledges the quaternary structure? Below please find a link to a file generated using pdb.gz structure data from Unwin model + a set of corresponding FASTA sequences, and attached picture shows the problem (revealed when alignment is set as shown in the picture, by clicking on the same 'boxes' as in the image). Long story short, I need to get the A subunit to complete the alignment in a different spot than it is automatically placed in, as it occurs in the whole structure twice.