For some reason, the pdb decided in the latest round of remediation to take a lot of short peptides and turn them into het atoms. For our example, we'll use 2tlx, where the dipeptide of Val-Lys used to be a chain B, but is now 2 het atoms.
When you open this pdb and look at these residues using the attached pml, only the peptide bond shows up. This is because the pdb decides to put in a connect record for the peptide bond to let you know it's a dipeptide, but leaves out the other connect records, figuring you know they're amino acids. It is ugly in pymol.
This may be just as much (or more) a bug in the files being created by the PDB, but I'm looking at these files in pymol and getting annoyed.