I have a PDB file with missing atoms and residues. I would like to build
the missing atoms only without touching the rest.
The easiest way to do this is to use the complete_pdb script, then write
out the model. No need to build an alignment and do the whole
comparative modeling run, unless you need to optimize the coordinates of
the missing atoms:
It seems that Modeller built the missing atoms (4) but I am ending up
with a new model that has fewer atoms than the original one (1512 vs
1558) although they both have the same number of residues.
When you build a model in Modeller it builds a single conformation using
all heavy atoms. If atoms are missing in your input file, it adds them;
if there are extra atoms they are removed. These extra atoms could be
alternate conformations for some residues (with an alternate location
indicator in column 17 in the PDB file, as per
http://www.wwpdb.org/documentation/file-format-content/format33/sect9.html#ATOM),
extra atoms such as hydrogens, or perhaps modified residues. But you
would have to look in the file to be sure.