[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [modeller_usage] (My) problems with BLK

Francesco Pietra wrote:
> I posted to a wrong list of Modeller, so I got no answer. Here the
> latest from my efforts, trying with a single chain to simplify.
> My pdb file ends like:
> HETATM 3346  CLA BLK A 859      78.668  51.520  63.082  1.00 26.90         CLA

The name of the residue in the PDB does not matter to Modeller if you
use the BLK residue type in the alignment - it'll preserve whatever is
in your input PDB. So in most cases it makes more sense to leave the
original PDB names (the official atom name, residue name, and element
are all 'CL' according to PDB).

> Models are built as if they had good DOPE score, though the log file complains:
> iup2crm_280W> No topology library in memory or assigning a BLK residue.
>               Default CHARMM atom type assigned:  CLA -->  CT2
>               This message is written only for the first such atom.
> Although I am unfamiliar with CHARMM (I use Amber), I understand that
> my setting of BLK is incorrect, CLA is represented by a wrong atom
> type (CT2 stands for some carbon?). I tried variants, vainly.

Exactly - since atoms in BLK residue have no topology, Modeller does not
know their type, so it guesses from the atom name. CT2 is one of the
CHARMM carbon atom types (see modlib/top.lib). The only solution here is
to define the topology for this residue type yourself, as covered in the
FAQ. But that probably doesn't buy you much, since chloride ion is just
a point - all you'll get is a subtly different VDW radius. As for the
actual value of that radius, that is something that you will have to
determine for yourself. LJ well depth parameters are less important
since by default soft sphere is used anyway (which has the radius as its
only atom-specific parameter).

> What I am trying is to represent CLA  (chloride ion) as accurately as
> possible, even if I am aware that the effective spatial situation
> depends on the environment (in particular the balance between ionic
> and covalent bonding) so that any set of literature data for CLA will
> hardly fit accurately my situation.

Modeller does not use electrostatics by default anyway, so your ionic
bonding will hardly be accurate. (You could certainly add individual
electrostatic interactions as restraints, however.)

	Ben Webb, Modeller Caretaker
Modeller mail list: http://salilab.org/mailman/listinfo/modeller_usage