On Mon, Dec 19, 2011, Jeffrey Thalhammer wrote:
> I am architecting a research pipeline using Amber. According to the
> tutorial in the amber manual...
>
> ...[Most] PDB files needs some editing before it can be used by
> Amber...Although Amber tries hard to interpret files in PDB formats,
> it is typical to have to do some manual editing before proceeding. A
> general prescription is: "keep running the loadPdb step in 16LEaP (see
> step 2, below), and editing the PDB file, until there are no error
> messages."
I don't necessarily want to discourage you from creating some automated
scripts, but I still think it is good general advice to expect to have to
hand-edit PDB files. Here are some reasons:
1. Many files have a ".pdb" extension, but don't come from the RCSB. These
can have all kinds of funny things in them.
2. Many or most "real" pdb files have ligands, solvent molecules, co-factors,
etc. Unless you are sure that you can ignore these uniformly, knowing what is
there can be important.
3. LEaP does not a present deal correctly with MODEL statements in pdb files.
Hence, if you get an NMR structure with 20 models in it, you will get a mess.
(Of course, your pipeline might handle this case.)
4. PDB files sometimes have multiple conformers with parital occupancy. These
are sometimes, but not always, identified as alternate conformers. It's hard
for a program (or even a human) to know how to handle such cases.
5. The probem of protonation states has already been raised.
Of the above, point 2 is the real killer: the number (and varying
importance) of things other than plain amino acid or nucleotide sequences
suggests that an automatic program would need to know a lot about
chemistry and biology.
(Maybe Scott, Dan or others can relate some horror stories about trying to
automatically prep sets of protein/ligand combinations for docking studies.)
This is *not* to say that Amber could not do a better job with some common
things, but rather that this is a hard problem....
...dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Dec 20 2011 - 13:00:03 PST
