The sequence alignments provided are based on a manual structure alignment.
See the paper for details of the construction of the alignment.  

All alignment files conform to the standards for the given file type (i.e. they
can be used directly in other software).  The "Raw CE" file is the exact same
alignment, but it is provided in the format that is output by CE.  The reason 
that it is provided is:
1) It contains the start and end numbers for the PDB structures
2) It contains lowercase residues.  These were not resolved in the structures,
and had to be aligned based on sequence and conjecture.  Thus, they are somewhat
less reliable than the rest of the alignment (only a small amount of residues were
unresolved, and seldom in important regions).  These residues are included in the
other alignment files, but they are shown in uppercase.  This is done because in
most formats lower case letters in the alignment are not allowed.