Generating parameters/cif files for macrocyclic ligands

From: Joel Tyndall
Date: 7 February 2012 21:44

Hi folks,

 

I have an intriguing problem. I'm trying to generate a cif file for a macrocyclic peptide (of the likes in pdb1d4k). They are cyclic tripeptides units. I can generate a pdb or mol2 file easily. I have used PRODRG to generate a .cif file and Coot read thjis in nicely. However, as it is cyclic one cannot adjust the dihedral angles. I have previously done this using CNS where you can break the tricyclic peptide into residues and generate parameters to specify bonds/links between the residues (which allows this kind of movement). I can't come up with a way to do this  without using CNS. I have looked ta J-ligand which allows for one link "between" two separate residues which precludes a macrocycle. I have looked at sketcher within CCP4 which reads the pdb files but I don't believe this can be done here. Within Coot I can refine the whole ligand but not certain components.

 

Any suggestions greatly appreciated . ( I may stick to coot refinement with fixed atoms at this stage)

 

Regards

Joel

 

_________________________________

Joel Tyndall, PhD

----------
From: <Herman.Schreuder

Hi Joel,

 

The way I solved this problem was by generating a linear peptide and then connecting the ends using a LINK card in the header of the pdb.

 

Good luck!

Herman

----------
From: Garib N Murshudov

Hi Joel

Herman is right:

If you are refining cyclic peptides then the easiest way is to use link record linking C-terminus with N terminus. the name of the link should be TRANS. Here is an example:

LINK             ALA S  21                     ASN S   1                TRANS

It will force ALA 21 to be linked (with torsion, angles, planes, bonds etc) to ASN 1 of chain S.

This way you do not have to create description for large molecule. If you still want to create one molecule and you have mol2 file with coordinates then you can use libcheck to generate full dictionary using following commands

libcheck

file_mol <mol_file_name>

nodist y

It should generate fdescription. However I would prefer using link record. this way you keep amino acid names etc intact. If you amino acids are not among existing then you will need to create their description first and declare them peptide.

regards

Garib

Garib N Murshudov 

----------
From: Joel Tyndall

Hi Garib,

 

Thanks for that (and thanks Herman). How do I declare a non-natural amino acid a peptide? My ligand contains two peptidic cycles (non-N to C) where the side chains are cyclised. I think I'll be able to use several linbk records for the connections but the non-natural amino acid are complicating the issue

----------
From: <Herman.Schreuder

Hi Joe,

 

Non-natural amino acids and links etc. remain a moving target. Almost every year, refinement programs change the way these things are specified. Here is how I would do it:

1) search the protein data bank to see if your non-natural amino acid is already present somewhere.

If yes, use the same residue name and atom names. As it stands now, the residue is most likely already defined in the libraries that come with the programs.

2) If not, I would define the monomer as a peptide under _chem_comp_desc_level. See e.g. $CLIB/data/monomers/c/CGU.cif. This should work. The cif file itself can be generated with Grade, libchk or Prodrg.

3)  If this does not work, one can always add more LINK records to make the connections.

 

My two cents,

Herman