Here, I'm building the case where there's a one-base deletion from the wild type: 
WT:  TGCTCCGCATTCCCGACATTCCCCAGCAAG 
Mut: TGCTCCGCATTCCC-ACATTCCCCAGCAAG
     12345678901234-567890123456789
            SSSSSSS  SSSSSSS
Everything else is the same as the original sequence I simulated. I've built this sequence in Avogadro. 
For creating the aligned structures, I need to offset from the previous residue numbers. 
For the full-length sequence, the dna atoms are (chain A and resid 8 to 14) or (chain A and resid 17 to 23). 
The first selection stays the same, but the second becomes (chain A and resid 16 to 22) (marked with S above)