Summary: ab initio conformational comparisons to crystal structures



 All,
         A few days ago I posted a request for information on how to go about
 minimizing a crystal structure using ab initio methods in order to examine
 the energetics necessary to get the molecule of question into the crystal
 conformation.  Attached are the responses I received.  Many thanks to you all.
 XXXXXXXXX
 ORIGINAL POSTING:
 Howdy all --
         I have a rather unique problem.  I have a structure which was
 resolved from an electron density map.  I wanted to compare this structure
 to another but when I went to do so using ab initio methods, I found that
 the single point energy of this molecule was horrendously large and any
 attempts to do energetic geometry minimizations led to different structures
 (changed torsions and bond angles).  Further investigation led me to the
 conclusion that the main problem is that the errors in bond lengths (as
 small as they were) were drowning out the effects which I wanted to see the
 energetic changes for (torsional and angular changes).  I recently saw a
 paper about a topic similar to this (I think it was in the Journal of
 Computational Chemistry)...  If anyone has any references to work in this
 area or any ideas about how I can go about comparing these structures, I
 would appreciate it if you passed it on...  All information will be promptly
 summarized and resubmitted...
                 Thanks,
                         Ray Crawford
                                                                "Absence of
 proof is not
                                                                  proof of
 absence..."
                          iguana (- at -) one.net                       -Richard
 Levine
                         http://w3.one.net/~iguana             The Lost World
      Micheal Crichton
 ##########
 The problem is that crystal structure mostly is not the structure which
 is minimal by energy. I recall the paper of Gilson et al. in Proteins
 a few years ago. They calculated energies in crystal and minimal states
 these were quite different :)
 Igor
 ------------------------------------------------
 IGOR TSIGELNY Ph.D.
 University of California, San Diego
 itsigeln (- at -) ucsd.edu
 ############
 Dear Ray,
 I just read your recent posting to CCL regarding electron density map vs.
 ab initio bond minimization. Now I don't know, which system this is,
 but could it be, that the experimental geometry, valid for the bulk,
 indeed does not represent an energy minimum of the molecule in vacuo.
 Of course, to decide this, one would have to look into the inter-molecular
 forces possibly present in the solid. Have you considered using a 3D code,
 atomistic or ab initio to clarify this aspect? One clue to the fact, that
 this maight indeed be the case should come from the experimental structure:
 are there any parameters that seem awkward or chemicaly unlikely, indicating
 a molecule under strain? Any obvious intermolecular forces (like hydrogen
 bridge bonds)?
 Cheers, Lutz Ackermann
                <^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^>
                <    Dr. Lutz Ackermann                        >
                <    Research Assistent                        >
                <    The Royal Institution of Great Britain    >
                <    Davy-Faraday Research Laboratory          >
                <    21 Albemarle Street                       >
                <    London W1X 4BS                            >
                <    UK                                        >
                <                                              >
                <    phone: +44-171-409 2992 ext 424           >
                <    FAX:   +44-171-629 3569                   >
                <    e-mail:  lutz (- at -) ri.ac.uk                    >
                <    http://www.ri.ac.uk/DFRL/L.Ackermann      >
                 vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
 ########
 Hi!
 This might seem a bit thick - but cannot you define your crystal structue
 as a zmatrix (get xmol to read it in - say pdb) then write it out as a
 gaussian zmatrix.  Now using opt=zmat , you can define your bonds as
 variables and everything else as constants.  This should fix your system
 at a given conformation but optimise your bond lengths.
 I can suggest other ways - but that sounds aabout the most straight
 forward.  Le t me know if I have completely missed the point!
 Best wishes
 Alex
  -------------------------------------------------------------------
 |Alexander J Turner         |A.J.Turner (- at -) bath.ac.uk                  |
 |Post Graduate              |http://www.bath.ac.uk/~chpajt/home.html|
 |School of Chemistry        |+144 1225 8262826 ext 5137             |
 |University of Bath         |                                       |
 |Bath, Avon, U.K.           |Field: QM/MM modeling                  |
  -------------------------------------------------------------------
 ##########
 Ray,
 We've seen similar (bond lengths incompatibility) problems, albeit mostly
 with MM and semiempirical calculations, in the context of a study we published
 some time ago (Nicklaus, M.C., Wang, S., Driscoll, J.S., Milne, G.W.A.,
 "Conformational Changes of Small Molecules Binding to Proteins",
 Bioorg.Med.Chem. 3(4), 411-428, 1995).  These problems, and the way *we*
 solved them, are discussed briefly in this paper.  We've also done some
 ab initio calculations---not reported there---where we saw the same effect
 of errors in bond lengths posing a problem for the exploration of torsional
 and angular changes.  Hope this helps.
 Marc
 ------------------------------------------------------------------------
  Marc C. Nicklaus                        Lab. of Medicinal Chemistry
  e-mail: mn1 (- at -) helix.nih.gov               National Cancer Institute, NIH
  Phone:  (301) 402-3111                  Bldg 37, Rm 5B29
  Fax:    (301) 496-5839                  BETHESDA, MD 20892-4255    USA
          WWW:  http://www.nci.nih.gov/intra/lmch/MCNBIO.HTM
 ------------------------------------------------------------------------
 #########
         Dear Ray,
         Your problem is in no way unique, it happens to everybody who try
 crystal structures in ab initio optimizations.  There are a couple of ways
 you can do this.  First, the largest errors are for bonds to hydrogen, you
 could simply set them manually at better positions before starting the
 calculation (C-H ca 1.08Å, give C an idealized geometry like tetrahedron
 for sp3, X-H generally shorter than C-H, there are lit. values for most
 types).  This is easiest if you can generate a Z-matrix for your X-ray
 structure.
         My favourite method is to bring the molecule into a molecular
 mechanics program, restrain all heavy atoms and let the program optimize
 the hydrogens.  Alternatively, you may want to allow also the heavy atoms
 some movement, most MM programs can do some kind of "tethering" to
 allow
 small relaxations.
         The most expensive option would be to restrict movement of the
 heavy atoms in the ab initio program, and let it find optimum values for
 the hydrogen, before starting the full optimization.
         An added complication: there is no guarantee that the crystal
 geometry is a local minimum in vacuo, especially if you have systems with
 strong charges, but it should definitely have a low energy if you allow
 relaxation.
         Regards,
         Per-Ola Norrby
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  *  Per-Ola Norrby
  *  The Royal Danish School of Pharmacy, Dept. of Med. Chem.
  *  Universitetsparken 2, DK 2100 Copenhagen, Denmark
  *  tel. +45-35376777-506, +45-35370850    fax +45-35372209
  *  Internet: peon (- at -) medchem.dfh.dk, http://compchem.dfh.dk/
 ############
                 Thanks Again,
                         Ray Crawford
                         iguana (- at -) one.net