On Sat, Aug 2, 2014 at 8:36 AM, Alastair Fyfe <afyfe@soe.ucsc.edu> wrote:

I also ran into this recently and was puzzled about the correct PDB encoding. Using the same residue number with differing altloc, atom name and element fields seems straightforward but what should the residue name be ? In my experience not all PDB-processing software is happy with different residue names for the same residue number. Is there definitive PDB documentation for this case?

This does appear to be permitted by the PDB (in addition to Phenix, of course) - a good example is crambin (1CBN), which has a naturally occurring polymorphism:

ATOM    352  N  ASER A  22       4.909  12.659  -3.127  0.20  3.03           N  
ATOM    353  N  BSER A  22       4.909  12.659  -3.127  0.20  3.03           N  
ATOM    354  CA ASER A  22       6.035  13.459  -2.622  0.20  3.04           C  
ATOM    355  CA BSER A  22       6.035  13.459  -2.622  0.20  3.04           C  
ATOM    372  N  CPRO A  22       4.909  12.659  -3.127  0.60  3.03           N  
ATOM    373  CA CPRO A  22       6.035  13.459  -2.622  0.60  3.04           C

There are other examples of crambin that are numbered differently but this is presumably due to the preferences of the investigator and/or the limitations of the refinement software used.  A counter-example is 3QXB, which has mixed metal sites:

HETATM 9585 FE  A FE A 301      -6.546  23.487  57.436  0.66 26.72          FE  
HETATM 9586 MN  B MN A 302      -6.549  23.490  57.439  0.34 25.22          MN  

I would say that if software can't handle the first case (same residue number, different residue name) that is a bug, but there doesn't appear to be consensus on the "correct" way to do things.  My preference would be to keep the numbering the same but the PDB will accept either convention.


-Nat