Page 110 out of 225 pages in this book.
Tripos Bookshelf  >  QSAR  >  QSAR Theory  >  CoMFA Theory

4.3.1 Preparing for CoMFA

1. A molecular database is needed, containing at least one conformation of each of the molecules to be studied or predicted, all mutually aligned in a plausible orientation.

2. All molecules must include atomic charges, if, as is usual, electrostatic fields are included. Although CoMFA results are not usually very sensitive to how charges are calculated, the MNDO method is recommended for conjugated systems.

3. A region definition file must be provided. Generally it is best to let the program compute a region automatically, which is indicated by check the Create Automatically box in the Add New CoMFA Column Dialog (or specifying CREATE_AUTOMATICALLY when asked for the region during creation of a CoMFA column). This creates a single grid, overlapping all entered molecules and extended by at least 4 Å along all axes.

Alternatively, you can specify all the properties of the region. A region comprises any number of rectangular grids, each fixed in Cartesian space, and defined by its upper and lower boundaries and grid spacing. One box, typically the only one, has X, Y, and Z boundaries which exceed the maximum and minimum atomic coordinates from every molecule under study. A single point is a valid region. A probe atom, the atom whose interaction with a molecule determines the magnitude of the field values, is associated with each region.

4. If you have an Advanced CoMFA license, you can transform Tripos Standard steric and electrostatic potential fields in several ways. Indicator fields set all energies below the steric and electrostatic cutoffs to zero. Useful analyses are best obtained with cutoffs substantially lower than the 30 kcal/mol cutoffs usually employed in creating Tripos Standard fields -- 10 and 1 kcal/mol for steric and electrostatic fields, respectively, are good starting points. Parabolic fields square the magnitude of the corresponding Standard field at each grid point.

H-bond fields potentials are zero at sterically prohibited points (above the value of TAILOR SET COMFA STERIC_ENERGY_MAX). Sterically allowed grid points close to H-bond donor atoms are assigned a nominal potential equal to that of TAILOR SET COMFA HBOND_WT (typically set equal to STERIC_ENERGY_MAX) in the donor field; all other donor potentials are zero. If a grid point is sterically allowed and is not close to a donor atom but is close to an H-bond acceptor atom, it is assigned a nominal energy also equal to the value of TAILOR SET COMFA HBOND_WT. This scheme follows the conventions introduced by Bohacek and McMartin [Ref. 13], with "close" being defined for the donor and acceptor fields by the variables TAILOR SET LEAPFROG H_ACCEPT_RAD and H_DONOR_RAD, respectively. Once an H-bond CoMFA column has been created, the acceptor component is nominally a "steric" field type and the donor component is nominally an "electrostatic" field type. This becomes pertinent when setting scalings directly and when interpreting PLS results.

Which transformation to apply is controlled by the Field Class menu selection in the Add New CoMFA Column Dialog.

As an aid during manual definition of a region, an Å ruler and any resulting region may be displayed by View >>> Ruler (RULER mol_area ON) and MSS: QSAR Manage CoMFA >>> Regions (QSAR COMFA REGION DISPLAY).

5. At least one alignment rule must exist for each molecule. While you can analyze molecules "as is", just as they are aligned in the database, you can save many other named alignments for each molecule, a procedure which becomes increasingly helpful as CoMFA runs are repeated under different assumptions.

You can put an Conformer column into an Molecular Spreadsheet and instruct the MSS to use the conformations and alignments in this column instead of those in the database when calculating molecular properties. This is generally more convenient and powerful than storing named alignments. Control of what structure is used is accessed by MSS: Options >>> Conformer Column (TABLE CONFORMER). Within an SPL program, suitable SLNs for CoMFA can be generated by TABLE COLUMN_APPEND CONFORMER.

6. Occasionally you may want to try CoMFA with energy fields other than the standard ones. Some possibilities, accessed mostly as CoMFA Field options, include MOPAC (HOMO or LUMO), Average, imported results of external calculations, and manual editing. Note that functional columns of MSSs can be derived from CoMFA field columns as well as from ordinary scalar columns. Field-with-field, field-with-scalar, and field-with-constant functions are all supported. The result of any functional calculation involving a field is always another field. During creation of the CoMFA column, the CoMFA process can be instructed to use an external field rather than calculate a new one.

Note that "field" when it refers to a type of data manipulated by SYBYL/QSAR, has a more general meaning than implied so far. Any set of numbers associated with a particular region can potentially be manipulated as a field. Some examples of this more general usage of field are the QSAR equation itself, the term-by-term prediction by the equation for a given molecule, or transformations of the real field. MSS: QSAR Manage CoMFA >>> Fields (QSAR COMFA FIELD RETRIEVE and QSAR COMFA FIELD EXTRACT) produce these field entities.