Quantum Refinement

Introduction

Stepwise

1. Download the X-ray raw data and its PDB model structure from, e.g., www.pdb.org.
2. Obtain molecular topology file (mtf), topology, and param files for CNS:
   to build an mtf file: download "generate.inp" from CNS website(http://cns.csb.yale.edu/v1.2/), properly modify it then run cns using it as the input;
   for topology and param files: The HIC-Up server at Uppsala University, http://alpha2.bmc.uu.se/hicup/xdict.html, can give topology and dummy parameter files when given the coordinates of the hetero site.
3. Convert the structure factors (reflection data) into the CNS format. There are many tools to do this, for example, sf-convert.
4. CNS INPUT for quantum refinement:
   Obtain the input from the ChemShell example folder. One may also download the input template minimize.inp and modify it with appropriate data obtainable from the header of the PDB, but make sure to
   (1.) add the output filenames of the X-ray energy "mmen2" and gradient "force2", as done in the example file; these filenames are needed in the qmrefine.tcl.
   (2.) in the following field, use "coordinate_infile=cns.pdb" and "coordinates disp=comp @cns.pdb1" (for 10^(-4)-10^(-6) of the coordinates.)
   
   In case if you donwload a newer version of CNS, you can also download the new input template, or alternatively, just modify the version at the line "checkversion" to be the same as the version of the CNS executable. This is something particular about CNS. It is also a good idea to start with comparing with the CNS input file provided in the example.
5. To run CNS, one needs to set some environment parameters in the submit script:
   
   setenv CNS_SOLVE '/ns80th/nas/users/yhsiao/cns_solve_1.21/'
   setenv CNS_LIB $CNS_SOLVE/libraries
   setenv CNS_MODULE $CNS_SOLVE/modules
   setenv CNS_TOPPAR $CNS_LIB/toppar
   setenv CNS_XTALLIB $CNS_LIB/xtal
   setenv CNS_XRAYLIB $CNS_LIB/xray
   setenv CNS_XTALMODULE $CNS_MODULE/xtal
   
   Note: "CNS_SOLVE" depends on where CNS is installed.
   
   

------------------end of CNS preparation-----------------------------
6. Use the pdb file built in CNS above to build pdb in CHARMM to ensure that the coordinates are the same. Especially CNS has already built missing atoms and residues from the original pdb. To maintain the same residue numbers, use the "offset" command in CHARMM.
7. Pay attention to the atom type definition to avoid atoms being rebuilt causing inconsistent coordinates with CNS. For example, if the template PDB uses CD1 ILE, change it to CD ILE, also rename the C-terminus oxygens to OT1 and OT2.
8. Add hydrogens and relaxed them but keep the heavy ones fixed. For electrostatic quantum refinement one needs to solvate the protein and relax the water molecules, but the heavy atoms in protein should be held fixed.
9. If you move the protein coordinates to a different origin, translate them back to the original positions. Otherwise they may be outside the defined crystal unit cell and give meaningless CNS calculations.
   
   ------------------end of CHARMM preparation--------------------------
10. Quantum refinement can be done using either mechanical or electrostatic embedding with slight different input. Directories
    examples/hybrid/qmrefine/electrostatic, and
    examples/hybrid/qmrefine/non-electrostatic
    contain working examples that can be of help.
11. Use smaller maxstep during optimization and larger convergence tolerance, as the mixed gradient (QM/MM+X-ray) seems very sensitive and harder to arrive convergence.
12. Construct converting tables "qmmm2xray.dat" and "xray2qmmm.dat" by using contrib/CNStoCh.f90. You only need to do this once when setting up a system. here you need to have an initial PDB file containing hydrogens. These two files are for converting the atmoic order between CNS and CHARMM (DL_POLY) and to get the gradients and new PDB in correct order, based on $CNS_TOPPAR/protein.top and CHARMM22. If different topology definitions are chosen, different order is followed thus the program needs modification. There might be problems for 2nd chain and onwards. Check if the order is correct before starting the refinement.
    
    The four number on top of "xray2qmmm.dat" are
    
    number of atoms in CNS (i.e., heavy atoms)
    number of atoms with hydrogen atoms added
    number of atoms in the solvation sphere
    number of atoms in the alternative conformation
    
    The last number is useful if more than one conformer will be included in one quantum refinement AND only the conformer of interest is built in CHARMM. However, only one of the conformers gets refined at a time while the others are kept fixed (only copying the coordinates at each step). Keep the fixed conformer ahead of the conformer to be refined, as the updated coordinates are concatenated onto the fixed part (filename defined by command altconf and altconf1). Please consult the example of multiple-conformer quantum refinement for details.
13. In the working directory there should be:
    X-ray raw data (CNS format)
    initial CNS pdb file
    CNS mtf file
    qmmm2xray.dat and xray2qmmm.dat
    CNS top and param files
    the usual parts for QM/MM
    
    ------------------end of QM/MM-based refinement setup----------------

Command Line Arguments

Under hybrid: