*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Dear all
Thanks for your many suggestions. These included partial disorder in
the ligand, incorrect chirality, bond cleavage, D/L isomerisation of
the peptide ligand, incorrect assignment of atom type and model bias
during building.
It turns out the solution was the following, from Corey Strickland:
Dear all
I have a problem with the ligand in my protein-ligand structure: it's
a peptide analogue and one of the peptide bond N atoms shows a very
strong negative peak (<-10sigma) in the Fobs-Fcalc map.
I am nearing the end of refinement and everything looks fine apart
from this one atom. Resolution = 43.0 - 2.18A, Rcryst = 19.3% Rfree
= 24.6%. I've been using CNS and O. I added the ligand as follows:
build with ProDrg, place with Arp/warp in CCP4i using default settings
and then some manual torsioning in O. I was careful to keep the
peptide bonds planar. The Fobs-Fcalc and map looks fine for the whole
ligand except for this one N. It also fits the composite omit map
nicely. The ligand B-factors are ~35 - which seems a little high
perhaps but not enough to warrant what I'm seeing. (I think occupancy
is probably not 100%, although I haven't tried refining occupancies).
The B-factor of the N in question is 57 - the highest in the ligand.
I've done 3 build-refine steps since adding the ligand and the
negative density hasn't changed.. Admittedly, these were limited to
minor adjustments and I didn't carry out more than15 minimisation
steps per cycle, however it doesn't look like the problem will be
solved by further refinement. There is no positive density that might
indicate incorrect positioning of the atom.
I asked the organic chemist who synthesised this ligand and he assures
me there is no way the atom could be anything other than a N. My only
other thought is perhaps this atom is different, in terms of
electrons, from a typical peptide N? It's the peptide bond before
the ligand C-terminus. Also the ligand is not a true peptide: the N
of the previous peptide bond (to the N-terminal side of the
problematic one) has been replaced by a C and the adjacent carbonyl
group has formed a gem-diol at the active site zinc ion. The
offending N and the gem-diol are separated by two CH2's and a carbonyl
group.
Can anyone shed some light? Or at least can anyone suggest something
else I could try?
Jean Watermeyer
Jean,
In your topology and parameter files for CNS, what is the atom
type given for the nitrogen (Also, check the atom name in the PDB file).
This sounds like the scattering factor assigned to the nitrogen is not
actually nitrogen. You can check the log for the cns run and look at
the list for the scattering factors that are used.
Corey
Hi Corey
Here are the lines from the top file that refer to the N in question
(which is called 'N' in the pdb):
MASS NB 14.0067 ! equivalent to N
ATOM N TYPE=NB CHARGE= 0.145 END
These are the only lines in the log files that refer to scattering:
XRAY> @CNS_XRAYLIB:scatter.lib
ASSFIL: file /usr/local/CNS/cns_solve_1.1/libraries/xray/scatter.lib
opened.
XRAY>! file libraries/xray/scatter.lib
The entry in scatter.lib for N is:
SCATter ( chemical N* and not
( chem "Ne" or chem "NE" or
chem "Na" or chem "NA" or chem "Na+1" or chem "NA+1" or
chem "Ni" or chem "NI" or chem "Ni+2" or chem "NI+2" or chem
"Ni+3" or chem "NI+3" or
chem "Nb" or chem "NB" or chem "Nb+3" or chem "NB+3" or chem
"Nb+5" or chem "NB+5" or
chem "Nd" or chem "ND" or chem "Nd+3" or chem "ND+3" or
chem "Npp" or chem "NPP" or chem "Np+3" or chem "NP+3" or
chem "Np+4" or chem "NP+4" or
chem "Np+6" or chem "NP+6" ))
12.2126 0.0057 3.1322 9.8933 2.0125 28.9975 1.1663 0.5826 -11.529
- and as far as I can tell the atom has been assigned chemical type N,
which should be right shouldn't it?
Can you spot a mistake here?
Thanks for your help
Jean
Jean,
Your problem is below. The scatter.lib entry says to select N
if it is not one of the others, i.e. NB. The atom type NB is reserved
for another element, Nb (Niobium) which is much heavier than Nitrogen!
Change the NB atom type to something that is not in the list below and
your peak will go away. Be careful, you will have to make other changes
when you change that 1 atom type, because there are parameters that are
defined by atom type, and the NB's in those statements will also have to
be changed.
This is a major hassle when you use other ligands than standard
proteins. I make my own copy of the scatter.lib file from CNS and just
include the normal elements. It is unlikely we will refine a structure
with a Nb element.
Corey
Dear Jean,
Check if the atom types are properly defined in topology and
parameter files. There is a possiblity that the contributing structure
factor will be high for this atom. One possibility : for example - if you
have mentioned NA instead of N in the defenition for toplogy and parameter
file..(which might mean that it is Sodium atom rather than Nitrogen as
defined in the scatter.lib).
HTH,
best wishes,
Natesh
IGBMC
I followed his advice and made a local copy of scatter.lib, from which I
deleted all the unusual atoms. I then reran the previous round of
refinement, and saw a significant decrease in R and Rfree as early as
step 1. (You also have to edit the later parts of the CNS script to
read in scatter.lib from a new location). I made a new Fobs-Fcalc map
and this looks like a perfectly normal ligand that still requires some
adjusting: no strange negative peaks anymore.
Robert Immormino suggests the follwing for getting around having to edit
the scatter.lib.
Jean,
I think you and Corey have found the problem!
Prodrg does give weird atom names...And if they conflict with heavy
atoms, then as you've found it can cause problems. The general
workaround that I have used is to draw the ligand, and submit that to
prodrg. Then take the output .pdb file and change the names of the
atoms and re-run prodrg with this .pdb. This gets rid of 99% of the
problems, but manual editing of the parameter and topology files is
still sometimes necessary. It might also be worth trying this
approach in parallel with changing the scattering factor library.
Best Wishes,
-bob
Thanks again. It's great to solve a problem so quickly!
Jean
