Re: [modeller_usage] Re: Ligand clues and water blues
To: "Mohsen, Al-Walid" <>
Subject: Re: [modeller_usage] Re: Ligand clues and water blues
From: Modeller Caretaker <>
Date: Fri, 06 Jan 2006 11:56:11 -0800
Cc: jo hanna <>,
Mohsen, Al-Walid wrote:
You have two messages and I am not sure where you are. I mentioned earlier
in my reply to replace the /...* with XXX* for the template and ...* for the
model in your alignment file.
'X', strictly speaking, is the UNK PDB residue type, which is probably
not what you want here.
model in your alignment file. It seems like you're still including the '/'
character, which is unnecessary at the end of the sequence and may cause
problems (shifts). The '/' is necessary to break the peptide sequence in the
middle of the sequence, not at the end. IF you do a dimer for example you
need it there between the monomers, but NOT at the end.
The '/' is a chain break character, so should be used if your ligand (or
water) is in a different chain to your protein (i.e. the PDB file has a
TER record between them). This also instructs the Modeller optimizer not
to complain if the distance between adjacent residues is too large
(since if they are in different chains, they can be far apart).
You MUST check the alignment file for shifts after you exe or start the
Modeler_Run AND the .ini file is written. Modeler 7 is known for introducing
'-' into the sequence and things get messed up. This occurs near the '/'
character.
This is not correct. Modeller has to introduce gaps into the sequence to
correspond with chain breaks in aligned sequences, since the '/'
character is not a real residue type, and so cannot align with anything
other than a gap. Modeller 7 does indeed put the - characters in the
wrong place when writing out the alignment files, but the internal
representation used for modeling is correct. (Modeller 6 did not have
this problem, because it did not handle multi-chain sequences properly
anyway. But you shouldn't be using Modeller 7 - use Modeller 8 instead,
which will read your scripts just fine, and which has this minor output
bug fixed.) Neither Modeller 7 nor 8 will overwrite your original
alignment file, in any case.
Everyone else can correct me, if I am wrong, but I do not see the problem
with water coming up as BLK. As long as it's in the proper position in the
3D structure, it's OK. Just replace BLK with HOH manually in the pdb file.
If you do the XXX aligned with ... then all HETATM including water are
treated as HETATM and renamed BLK.
The difference is a minor one; BLK residues are treated as rigid bodies,
and are not modeled by the energy function (although simple harmonic
distance restraints are used to keep them near approximately the same
set of residues as in the template). Water residues are treated in the
same way as ordinary amino acids, although by default Modeller does not
model hydrogen atoms, so each water will be represented simply by the O
atom.
I would leave water_IO = on out. Do not assume that all the commands work
perfectly right. As I mentioned above Modeler 7 has this shift issues that
was not present in Modeler 6.
I am not aware of any release of Modeller which does something incorrect
with water_io. It simply instructs the code to read any residue from the
PDB which matches the water types defined in modlib/restyp.lib (i.e.
HOH, H2O, OH2, MOH, and WAT).