[phenixbb] [ccp4bb] Ligandfit - problem with ligand_start

Terwilliger, Thomas C terwilliger at lanl.gov
Mon Nov 25 08:24:24 PST 2013


Hi Danilo,

The most likely problem is that the ligand_start file has to have enough atoms in it to define a fixed group in the ligand.  The way that ligandfit works is that it breaks a ligand up into rigid groups connected by rotatable bonds.  The ligand_start file has to contain one complete rigid group.  So probably if you supply a little more of the ligand then it might run.

Just to test this, can you try this test of using a ligand_start file:

  phenix_regression.wizards.test_resolve resolve test_fit_start

And then if that gives an "OK", try this:

cd test_fit_start
phenix.ligandfit ligand=side.pdb ligand_start=start.pdb resolve_map.mtz quick=true

Does that run?  If you the look at 

  LigandFit_run_1_/TEMP0/resolve_ligand_fit.log

Does that say 

  >ligand_start start.pdb
Starting configuration of ligand will be read from: start.pdb

If all that works, then try adding more atoms to your ligand_start file.

Let me know if that doesn't do it!
All the best,
Tom T

  
  Dear all,

  I am working on a membrane protein covalently bound to a molecular
 antanna: it is known that this molecule binds to lysine residue but I do
 not know how many and which lysine residues it binds. 20 diffraction
 datasets of this protein-ligand complex have been obtained and now, I
 would quickly localize the ligand using the Fo-Fc map of each data set
 and using the information on the covalent bound protein-ligand.

  Ligandfit tool (PHENIX) seems to be indicated to do this; to use the
 information on the covalent bound, I am using the ligand_start keyword
 with a pdb containing a ghost atom (however present in ligand model)
 perfectly superposed to the lysine atom that should bind the ligand.

  The command used is:

  phenix.ligandfit data=prot.mtz model=prot.pdb ligand=lig.pdb
 ligand_start=lig_start.pdb  input_labels="FOFCWT PHFOFCWT" \
 refine_ligand=True  \ nproc=32 \ cif_def_file_list=lig.cif

  description:
  - prot.mtz       (data)
  - prot.pdb       (protein without ligand)
  - lig.pdb        (ligand containing ghost atom)
  - lig_start.pdb  (ghost atom superpose to NZ of a lysine)
  - lig.cif       (restrain of lig.pdb)

  Strangely, no ligand is found at the end of the process even reducing
 ligand_cc_min to 0.01. I have run the same command by using an other
 protein where an other ligand has been correctly fitted but, also in
 this case, no ligand has been detected. Conversely, without the use of
 ligand_start, ligandfit properly localizes the ligand.

  I'm doing some mistake in the use of ligand_start? Do you know an
 other tool to perform a ligand fitting in these conditions?

  Thanks for your answers.

  Danilo


More information about the phenixbb mailing list