Site Searches
There are a number of changes relating to virtual screening.
Hit Refinement
The most significant change for virtual screening is a replacement refinement routine which more robustly adjusts the position of each conformer which fits to the pharmacophore within the active site in order to minimise the score. Inevitably, this is also slower than the initial implementation (although attempts to restore the speed are underway).
Trace and Profiling
New values have been added for ITRACE which provide for a more comprehensive analysis of where the time is spent and why hits are not found for molecules. A trace value of 1 is especially useful when performing validation studies to reproduce how known actives interact at a receptor site. For searches which take an excessive time, 1024 provides a more detailed progress reports and (on completion) a breakdown of the time spent in key algorithms. Possible ITRACE values are summarised in the THINK Theory Manual.
MODIFY RATIO=R
A keyword has been added to the MODIFY command which means that it is no longer necessary to set the CPK to VdW ratio prior to issuing the MODIFY INTERACTIONS command (only to restore the original value later).
Pharmacophore Scores
The required group implementation has been extended to use group numbers to compute a score for each pharmacophore when the site points are created with a bound ligand present specified using the LIGAND keyword. This extension alleviates the need to edit the query file in order to eliminate site points which are remote from the site (although in most cases we recommend you do so). In essence, group numbers 1-4 are assigned depending on the proximity of the site point to the ligand:
| Group number | Description |
| 1 | Site point overlaps with ligand (the separation distance is less than 1 Angstrom) |
| 2 | Site point in contact with ligand (the separation distance is less than using a radius of 1.0 + VdW radius of a ligand atom) |
| 3 | Site point is less than DMNLOS (6 Å) from at least 3 non-hydrogen ligand atoms and has an unobscured 'line-of-sight' to them (ie no protein atoms in a cylinder of radius 3.2Å) |
| 4 | Any other site point |
Output files
If only the best conformer geometry was output (ISTALL=0), the serial and group numbers used to correspond to the last solution found whereas they now correspond to the best solution. In addition, for 3D and SITE searches without a protein present, the serial numbers are now set.
For conjugated systems, some different results may be observed as the test to determine whether a conjugated bond was too crowded has been relaxed slightly.
Screen Saver Clients
It is now possible to configure the screen-saver to display the THINK icon rather than the molecule being searched. This is important when the molecules being searched are proprietary and their structures should not be displayed.
An alternative to the use of environment variables has been added which requires the file think.env to be located in the same directory as the executable. This is especially valuable for screen saver clients, when creating environment variables is dependent upon the operating system. The format of the file is as follows:
THINK_EXEC:=z:\programs\think\v113\
The implementation uses THINK symbols which provides a means of (re)defining such variables (note that the symbol name ends with a colon).
The logfile for screen-saver clients is now opened in APPEND mode so that information from previous executions of THINK is now retained.
PDB Files
THINK will now automatically create separate a separate molecule for all the water molecules and for each ligand providing:
In the event of NAME records being encountered then the atoms are placed in these molecules (and any associated CONECT records must be moved to before a NAME record for another molecule). This approach is recommended when ligand binding includes one or more water molecules which should then be included in the protein molecule. It is also still necessary to edit a PDB file to insert NAME records when the ligand is a peptide (for which there will not be a HET record). The name of the ligand will be taken from the HETNAM record unless a HETSYN record follows.
THINK continues to extract the molecule name from COMPND records but will now use the MOLECULE, SYNONYM or EC subrecords (if present) or columns 63:66 of a HEADER record (if non-blank) which is reserved for the PDB code.
When generating connectivity for amino acid chains, THINK now checks that the distance between the terminal C of one amino acid and the starting N of the next are within a distance of 2Å.
An extension to PDB files is now used to read and write the group number from columns 67:69 (which are unused). This is important to support the scoring of pharmacophores based on the group numbers of the site points.
Command Scripts and Symbols
Interrupts
When interrupting a command script (for example entering control_C in the command line or clicking CANCEL), all script files are closed and command is returned to the keyboard.
Error handling
When an error is detected during the execution of a command script (and no interrupt is generated) the behaviour has been changed to jump to the label ON_ERROR. This may also be a symbol in order to jump to different labels in various parts of a script depending on the translation of the symbol. In addition, the symbol ON_ERROR may have the value CONTINUE to indicate that execution should continue with the next line in the script (ie the error should be ignored).
Reading and writing from and to files
READ and WRITE commands have been added to support reading and writing of text from and to files using the following syntax:
READ filename SYMBOL
WRITE filename SYMBOL
where SYMBOL is the name of a symbol containing the text. CONSOLE and LOGFILE are reserved values of filename used to write to the console window and the logfile respectively.
In the event of an attempt to read beyond the end of the file, a jump to the label FILE_END is executed (and an error condition is generated if this is undefined). If an error condition is detected during the read or write a jump to FILE_ERROR is executed. Analogously to ON_ERROR, FILE_END and FILE_ERROR may be symbols.
Symbols
It is now possible to extract (but not set) substrings of symbols eg TEXT(3:5). These may be combined with array indexes, use symbols, be omitted etc for example
LINES(2)(J1:J2)
LINES(J)(J1:) or LINES(J)(J1:*)
The implementation of local symbols has also been improved to support parameters to logfiles. Within the logfile these may be referenced by Pn where n is an integer in the range 1 to 9.
Graphics
It is now possible to use dashed lines to display interactions or contacts between two molecules in addition to atom mapping using the OPTIONS keyword of the DISPLAY command and QUERY keyword to specify the second molecule. This is normally used to display which atoms in a 3D molecule are mapped by serial number to the query (or in the case of SITE search, to the protein). It can also be useful to visualise CPK contacts and/or interactions between a docked molecule and a protein site for the results of SITE searches (or for any ligand-protein crystal structure). The interactions are a subset of the CPK contacts for which both atoms are centres (eg hydrogen bond donors, acceptors etc). In order to display hydrogen bonds it is necessary to increase the CPK to VdW ratio to about 1.25 and disable the check for attached hydrogens in proteins (because the hydrogens are normally omitted) using the following command
CUSTOMISE CPK=1.25 DONOR=NOHYDROGEN
Some small changes have been made to the interactive manipulation capabilities, including preventing re-scaling occurring following motion and the implementation of the shift key to allow a molecule to be scaled.
Miscellaneous
Logfiles
The think logfile (think.log) normally includes all error messages. This has now been extended to include the message stack when it is visualised using the MORE button on the message window.
Property Calculations
Some new property calculations may be observed including RMS, RMS-NoH, RMS-All and Holes and the addition of the QUERY keyword. These are currently under development and their use is not documented or supported.
SD Files
Some small changes have been made to enable hydrogens to be correctly added
to 3D files from which they were absent. In addition, it is hoped that a subtle
change to the order of atoms on the bond record will allow some other software
to read the files without problems.