No, because you already know in advance that they differ in ways that are going to affect the results. ExbB codes for three membrane segments, motA codes for four. Right off the bat that's going to skew the percentage of correspondences in a way that really isn't particularly revealing or useful.
Essentially, you're putting two different fruit baskets side by side and trying to compare them. You're much better off if you break them down into their smallest functional subunits, and then compare those. That way, you can pull out an apple from each basket and compare them directly, rather than comparing the bunch and watching the correspondences get lost in the noise created by the fact that one basket has a banana where the other has a pear, and the second has two oranges where the first has none. Apples to apples ;)
What? That is the purpose of measuring. Then an exact correspondance of the functional part means they are the same.
Apples to Apples and Fruit baskets to fruit baskets.