Chapter 14. Conformational Searching

 

            GLOBAL-MMX (GMMX) is a steric energy minimization program which uses the currently selected force field to search for the global energy minimum and other low energy local minima.  Processing by GMMX is done in two stages.  The first stages randomly searches over the selected rings and rotatable bonds and keeps all conformers within 3.5 Kcal of the lowest energy conformer found (Eminim) during the minimization.  The second cycle reminimizes the structures found in the first cycle and keeps only those which are within 3.0 Kcal of the lowest energy conformer.  The default energy windows for the two cycles can be changed.  In addition to an output file containing the coordinates of the final structures, a textual summary file called '<OUTPUT_FILENAME>.pkm' is produced which lists the energy  and Boltzmann distribution of each low energy conformation.  This file may also include a listing of query operations (PMR coupling constants, distances, angles, and dihedrals) for each structure as well as a Boltzmann averaged summary.

 

            Although the conformational searching techniques that follow are unique in their approach, the methods described by the Still group (see M. Saunders, K.N. Houk, Y-D Wu, W.C. Still, M. Lipton, G. Chang, and W. Guida J. Am. Chem. Soc. 112, 1419 (1990) and references to previous work cited there in.) were the inspirational source for this work.  The search techniques in GMMX are based on the methods used in BAKMDL, developed by Professor Kosta Steliou of Boston University, and ported to the MMX force field by Mark Midland of UC Riverside and Joe Gajewski and Kevin Gilbert of Indiana University.  We wish to thank Professors Still and Steliou for sharing with us unpublished work and code that greatly enhanced our routines.

            Conformational space can be searched by GMMX in three ways.  These are the Mixed method, the Bonds method and the Cartesian method.  The Bonds method randomly selects a subset of the bonds designated by the user for rotation.  Bond rotation can cause large changes in the shape of a molecule.  The Cartesian method randomly moves a subset of all heavy and nonvolatile atoms in 3d space causing small changes in the shape of the molecule.  The Mixed method alternates between the Bonds and Cartesian methods and is the preferred method.  This procedure is especially efficient with cyclic structures having side chains.  You may select ring bonds for rotation as well as the side chain bonds but restricting the bond selection to the side chain bonds usually is all that is necessary if only 3 to 7 member rings are present.  The coordinate (Cartesians) movements will apply to the entire structure.

 

 

 

Figure. GMMX setup dialog box.

 

            When GMMX is first selected from the Analyze Menu a dialog box will be presented as shown above. This dialog box controls the setup and running of GMMX.  Items that can be set directly in this dialog include the jobname, the search method (bonds, coordinates or both (default)), and the energy windows for the first and second cycles.  The buttons Setup Rings, Setup Bonds, Comparison Method, Setup Queries, Options, and Read Job all call up further dialog boxes to setup the various options.  Save Job allows the user to save a file of the current GMMX options, and Read Job reads this file.  Run GMMX starts the first cycle processing while Run 2nd Cycle starts the job in the second cycle, which presumes that a file of structures saved from the first cycle is available.

 

Setup Rings

            GMMX will automatically search for all rings in the current structure and the results will be listed in a scrolling list window. To search on a particular ring select that ring from the list of rings and the select the Add button.  In molecules containing fused or multicyclic ring systems all possible permutations of the rings will be listed.  Thus trans decalin will have entries for two six membered rings and one ten membered ring.  Care should be taken in selecting the correct ring entry to search.  During the search the ring will be broken between the first two atoms of the Atom Number list.  These are displayed as Closure Atom 1 and 2.  The Closure Atoms may be edited but they must be adjacent atoms.

 

 

 

 

 

 

 

 

 

 

 

 

Setup Bonds

            To designate rotatable bonds select the Setup Bonds option in the dialog box.  A new dialog box will be shown.  The atom numbers of the rotatable bond can be entered in the edit boxes.  When the ADD button is selected the atom numbers are read and checked to see that a bond exists between the two atoms.  If a bond is found then the default rotation increment for this bond is then determined and the bond is added to the scrolling list of bonds for searching.  Bonds may also be entered by using Sel-Bnd from the Draw Tools.  To use this option, select the bonds before bringing up the Setup Bonds dialog box.  The selection may be done while the main GMMX dialog box is up.  Just move the box to the side and rotate and select the bonds.  When Setup Bonds is selected, the selected bonds will appear in the scroll list.  To delete a bond, select the bond from the scrolling list and then select Delete.

 

 

Bond Resolution

            If the bond is an ester or an amide, the resolution set will have two angles (0 and 180 degrees).  If the bond is an SP3-SP3 bond, then three angles are used (0, 120, and 240 degrees).  If the bond is SP3-SP2, twelve angles are considered (0, 30, 60, 90, 120, 150, 180, 210, 240, 270, 300, and 330).  If the bond is SP2-SP2, 6 angles are chosen (0, 60, 120, 180, 240, and 300).  Finally, if the bond is a ring bond, then if the ring is 10 atoms or less, the resolution set will have 12 angles of 30-degree increments as above and if it is 11 atoms or more, 6 angles of 60-degree increments will be used.   The user can override these default settings (except for no-rotatable bonds) and enter their own values.

 

Additional Ring Information

            For each ring defined, GMMX will calculate closure angles and a closure window.  Again, the user can accept these values or enter others.  Within GMMX each bond is randomly rotated by each angle in the resolution set.  The newly created geometry is then analyzed to see if each ring can be reclosed within the defined limits of closure.  It is checked for bad 1,5 C-C VDW  interactions and trans annular interactions.   Structures with up to 2 additional bad 1,5 C-C VDW interactions beyond the original structure may be kept.  The structure is also screened for epimerizations and all other constraint criteria such as distances, dihedral angles etc. imposed during the interactive procedure.  If all the constraint criteria are satisfied, the conformer is processed by GMMX.

 

Setup Query

            The Setup Queries dialog box allows one to specify query information for the second cycle.  Clicking on the four options will bring up dialog boxes for entering atom numbers.  For distances the atom numbers must be entered in sets of 2, for angles sets of 3, for dihedrals sets of 4 and for coupling constants sets of 2.  This process may be automated by performing the query on the structure on the screen.  This may be done prior to setting up GMMX or while the main GMMX dialog box is on the screen.  Use Query on the Draw Tools.  When one of the four options is selected, the query information will appear in a scroll list.  The final output file, jobname2.pkm, will contain a list of the queries for each structure as well as a summary total based on a Boltzmann average.

 

 

 

 

 

Comparison Method

 

            The Comparison Method dialog box allows one to choose various options for structural comparisons during the search.  The default is to compare all nonvolatile atoms.  Nonvolatile atoms are those that are not removed in an H/AD operation, i.e. all heavy atoms and hydrogens on O or N.  The equivalent atoms and Specific Sequence options have not yet been implemented.  Numerical isomers are found in unsubstituted rings such as cyclooctane.  Reflection isomers are found in unsubstituted acyclic compounds such as butane.  A set of specific atoms may be compared by clicking the Specific non-volatile atoms box and entering a set of numbers in the edit box.  The default comparison routine uses a fast analytical method during the first cycle and a slower but more accurate method during the second run.  It is usually better to err on the side of over collecting on the first run and then weed out duplicates on the second run.

 

 

 

 

 

 

 

 

Options

 

            The Options dialog box allows one to change several options for the search.  Chirality will be checked and enantiomers rejected if the box is checked.  Hydrogen bonding will be on by default.  If the system has pi atoms turned on, a pi calculation will be performed.  Note that this may not be necessary and may require more time for rigid systems like benzene.  By default, structures will be screened for bad 1,5 interactions and rejected if found prior to minimization.  Conformational enantiomers such as the two gauche forms of butane will be kept.  An early bail out of the minimization can be done.  This applies a sliding high energy cut off of 5-15 kcal/mol above the minimum energy found to the structure during the 50-100 iterations of the first-derivative minimization.  If the structure is 15+ kcal/mol above the lowest energy structure at the 50th iteration, it will be rejected and the search will move on.  This cut off is not used during the second cycle.  This method may lead to premature bail out if there are a lot of hydrogen bonds in the structure.  The structures will be sorted by energy.  The option to sort by heat of formation has not been implemented.  The Boltzmann temperature is used to calculate the distribution of isomers.  The cutoff distance and energy are used during structure comparisons.  During both cycles all structures within 0.25 kcal/mol of the current structure will be examined.  During the first cycle a structure will be considered to be unique if any atom is more than 0.25 Angstroms away for the least squares fit.  In the second cycle the RMS average must be greater than 0.25 Angstroms.  The random number generator uses the seed value.  A search starting with the same starting structure will always follow the same path unless the seed value is changed.  The search will stop after the lowest energy structure (emin) has been found 5 times or there are 25 duplicates in a row without a new structure.  However at least 50 structures have to be minimized.  The default number of duplicates required is 2.5 times the number of heavy atoms with a maximum of 50.  The search will also stop if 100,000 structures have been minimized.

 

Run GMMX

            Once all the setup is done GMMX can be run. In all methods the original structure is minimized to generate the initial minimum energy conformation. The structure is then modified by either the bonds or Cartesian method and the new conformation is checked for bad-1, 5 interaction, and epimerization.  If all the acceptance criteria are satisfied, the structure is minimized.  If the minimization yields an energy acceptable conformer that HAS NOT BEEN FOUND BEFORE, it is saved and its geometry is used to spawn a new conformer, otherwise, it is rejected.  Only accepted conformers are permitted to carry-on the search.  If attempts to create a new conformer fail, based on the acceptance criteria, it is sent back to try again until a NEW ACCEPTABLE conformer is created. Statistical searching on coordinates should be selected whenever a rigid or fused polycyclic system is to be analyzed.  If there are side chains then the Mixed procedure should be favored.  The Bonds method is best for acyclic structures.

 

Run 2nd Cycle

            Run GMMX will automatically run the first and second cycle of the search.  The second cycle refines the structures found in the first cycle by throwing out unnecessary duplicates and cutting down the energy window.  It also performs any Query operations.  The structure output files from both searches are kept.  You can send the search output through the second cycle at a later time by using this option.  The run will use any criteria (such as Query information) which you have selected.

 

 

 

 

 

 

 

Save Job

            This option will bring up a file-save dialog box.  The default file name will be the Job Name entered in the GMMX dialog box.  The file save will save <JobName>.pcm and <JobName>.inp files.  The inp file contains the name of the pcm structure file and the directions for doing the search.  Thus the pcm file cannot be deleted or renamed if you wish to run this job later.  The inp file may be read by the DOS or text-based version of GMMX.

 

Read Job

            This option brings up a file read dialog box and reads the inp file saved in Save Job.  All options will be updated to the settings in the job file.