Release Notes v1.20

OpenGL

The graphics implementation has been extended to support OpenGL. This provides cylinder, ball-and-stick and sphere displays as well a line display (similar to the Windows graphics). If the graphics card supports no more than 256 colours, or the THINK_OPENGL environment variable is set to 0, the standard windows display code is used instead. The display style can be controlled via the additional STYLE keyword in the DISPLAY command or from the Display dialogue. Alternatively, a pop-up menu displayed using the right-hand mouse button can be used in accordance with the usual window conventions.

The display of 3D molecules has been modified to accomodate a right-hand mouse button pop-up menu. Users can select the manipulation mode from this menu and the molecule will be dragged when the mouse is moved with the left mouse button held down. The old, less-standard tracking mode is still used if the environment variable THINK_TRACKING is set to 1. The implementation has been extended to include a scale option.

3D and Site Searching

If the results of 3D and site searches are stored in SMILES format, the serial numbers which map the hit to the pharmacophore atoms (and thence to the protein for a site search) are now included in the output. The extension to the SMILES format uses {n} where n is a number immediately preceding the atom in a similar way to chirality. THINK can also read such files.

An experimental enhancement to the scoring function has been added following the work of Protherics and Schrödinger. The view has been expressed that the ChemScore function incorrectly predicts the binding of some molecules which have lipophilic ligand atoms and hydrophilic protein atoms in close proximity (or vice versa). The function has been extended to include a term that penalises lipophilic-hydrophilic contacts in a similar manner to lipophilic-lipophilic contacts. The extra term uses the constant GSTBAD. By default this has a value of 0.0, so the term does not contribute to the score. The command "LET #GSTBAD = 0.058" would be used to set the constant to 0.058, which eliminates approximately half of the total number of hits in typical examples. To date, no extensive validation has been performed on the new term, nor have the other constants been re-tuned. Consequently, setting GSTBAD means that typical scores increase by 20kJ and some desirable hits are eliminated along with many undesirable hits.

R-group searching

A substantial change has been made to provide a more intuitive means of specifying both the required substructure in the reagent, and the location of the variable portion which becomes the R-group. For instance, when using SMILES, a query of [Z]N would locate all amines (but not ammonia since Z matches any atom except hydrogen). When this query is used for an R-group search in THINK 1.20, the NH₂ group is replaced by a generic connection [0]. The same approach can also be used when there is more than one variable part (or R-group) per molecule.

Minor enhancements

• The trace output has been extended to include the message "Pharmacophore not matched" during a site search when the substructure is matched but no pharmacophore is matched
  
  
• If a ^ character is specified at the start of a molecule name (like in an atom specification) it is now ignored
  
  
• The default value for RATIO in MODIFY INTERACTION is now 1.25
  
  
• The table calculated by "LIST INFO=PROPERTIES QUERY=query" (where query is the original ligand) now includes the additional molecule-query and molecule-protein fields (if the protein is present). The molecule-query fields include the RMS for the matched atoms and, providing the same number of atoms are present in the hit and the query, the RMS for non-hydrogen atoms and all atoms. These fields are useful when comparing crystallographic and predicted bound-ligand geometries. The ligand-protein fields include a breakdown of the constituent terms of the docking score.
  
  
• The spreadsheet display has been modified so that when the molecule name is clicked in addition to the 2-D display window being updated the 3-D display is also updated if it is open (and the molecule is in the list that can be scrolled in this window). This is useful when viewing ligand-protein complexes.
  
  
• The way in which the COMPND keyword in PDB files is used to provide molecule names has been revised. Details are provided in the THINK Theory Manual. Use of the nonstandard NAME keyword is still recommended for providing molecule names. If a standard PDB files containing HET records with associated HETNAM or HETSYN records is read, THINK will create separate molecules (providing they are not connected).
  
  
• When running in a distributed intranet or internet configuration using THINK as a screen-saver, the display style will randomly change between the new OpenGL options.

Bug fixes

• The processing of tautomeric groups in a protein could lead to some centres being incorrectly
  omitted from a query generated with MODIFY INTERACTION. This has been corrected.
  

Release Notes for v1.14