Release Notes for v1.13

Release Notes v1.12

This release was created primarily to integrate changes made to THINK to work within the United Devices platform. Some other enhancements have been made, especially to site search.

Enhancements for United Devices

The details of these code changes should not concern normal users and are described elsewhere. They were necessary for the CAN-DDO project (see www.chem.ox.ac.uk) to take advantage of the features of THINK v1.11, especially the increase in speed.

Site Search

Ligand position refinement

A Simplex minimiser has been implemented to adjust the position, orientation and conformation of the ligand within the active site in order to improve the estimated free energy of binding. Conformations which meet the geometric constraints and have an initial score below the value of GSTMIN (by default 1000) are refined. If the resulting conformer score is below the SEARCH CUTOFF value GSTHIT (by default 0), the conformer is considered a "hit". This enhancement is important when attempting to reproduce crystal ligand-protein binding geometry, although it is sometimes necessary to decrease the torsional increment by increasing the number of points eg:

CUSTOMISE SINGLE=12 ALPHA=12

The CUTOFF keyword for the SEARCH command now modifies GSTHIT whereas in the past it set GSTMIN. The results of previous versions can be reproduced by setting GSTHIT to a value equal to or greater than GSTMIN by use of the SEARCH CUTOFF keyword or using the command line:

LET #GSTMIN = 1000

For each cycle of the minimiser, a random vector is chosen in terms of the mix of X,Y and Z translations and rotations, and bond rotation initial shifts. A minimum is then found along this vector by repeatedly re-computing the free energy of binding whilst adjusting the length of the vector. It is possible to alter the maximum number of cycles or iterations by modifying the common variables NCYCMX or NITEMX respectively.

It is conceivable that the optimised ligand will no longer fit the pharmacophore (with the tolerances) unless it is constrained to do so. This was suggested by C A Baxter et al in JCICS 40.2 p254-262 (2000). In THINK, when the deviation of the matched atoms exceeds the tolerance there is an additional contribution to the free energy consisting of K(d-t)2, where K is the value of the common variable GSTRES (default 1000); d is the deviation and t is the tolerance. Setting GSTRES to 0 disables the constraint.

Conformational increments

The variable RCNFMX, when >0, is used to skip molecules that have more than this number of possible conformations, based on the current torsion increments. The implementation has been modified so that the torsion increments will be automatically decreased if necessary, in two passes over the rotating bonds

The list of rotating bonds is ordered according to the number of centres moved by each bond (in ascending order). The bonds are considered in turn until the revised value of the number of conformations falls below RCNFMX.

Required groups

Users can now select site search solutions which utilise pharmacophores that include a list of alternatives for each point in the pharmacophore. This can be important when molecules are required to interact with one or more residues. The implementation utilises the group number of the centres in the query, and the common variables IPHGP1, IPHGP2, IPHGP3 and IPHGP4. The atom group numbers are used as a bit mask. This means that the list of alternatives can overlap by using group numbers which correspond to two or more bits being set. Thus setting IPHGP1 to 4 (corresponding to bit 2 set) will require one pharmacophore point to be selected from the subset of query atoms that has a group number which sets bit 2 ie 4, 5, 6, 7, 12, 13 etc. If IPHGP2 was set to 8 then a second point would be selected from those atoms with group numbers 8, 9, 10, 11, 12, 13 etc.

At present, group numbers are stored using an extension within MDL SD files, and may not be read from PDB files. It is also possible to change the group number of an atom using the MODIFY command eg:

MODIFY CHANGE=(342)HDON GROUP=5

If there are group numbers in the query corresponding to bits 4, 5, 6 or 7, these implicitly set IPHGP1, IPHGP2, IPHGP3 and IPHGP4.

2-Centre pharmacophores

It is now possible to set the number of centres to 2 using the command:

CUSTOMISE CENTRES=2

The implementation uses an additional pair of points generated from the centroid of the molecule and the centroid of the complementary centres in the query. The 3 pairs are used to place the molecule in the active site (where its position and torsion angles may then be automatically adjusted using the new refinement option).

Aromatic complementary centres

The code underlying the MODIFY INTERACT command to create complementary centres from a protein has been extended to create aromatic centres. These are positioned above and below the plane of aromatic rings at a distance of 3.2Ǻ from the ring centroid.

Checkpoint file

In the event of an unexpected halt to the application (eg a power failure), it is now possible to restart a search from the point of failure. The implementation stores the progress made in a file “search.dat” in the current working directory. It is important to delete this file if another search (with the same query and file names) is required following a failure, otherwise the data in the checkpoint file will result in part of the search being skipped.

Obviously, in the event of a THINK software crash, restarting will again result in a crash when it processes the same molecule. Interrupting the search via Ctrl-C or Cancel results in the checkpoint file being deleted.

Trace output

It is now possible to record the destiny of the molecules being searched to the console window and the “think.log” log file. For a file containing a large number of molecules, this output can be voluminous and time-consuming. Consequently, the option should only be used when attempting to analyse why a molecule was not found. The common variable ITRACE is used as a bit mask and setting bit 0 (value 1) produces an additional line of output which indicates the following possibilities for a molecule

When a molecule is filtered, the field or substructure number is also reported.

An additional window can be enabled in order to monitor the processing of individual molecules by setting bit 10 (value 1024). This bit may be set in addition to other bits. This window then shows the main routine calls, the time that the routine was entered and the molecule being processed. In most cases, this should confirm that the software is working normally even if the time used is excessive.

Additions to CUSTOMISE

This release includes an increase in the number of settings which can be altered through the CUSTOMISE command, as summarised in the following table:

Keyword Purpose
TIMEOUT Set the conformational generation maximum time
TOLERANCE Set the minimum distance tolerance for 3D and SITE searches (the sum of the atom radii will be used if this is larger)
VOLUME To reject 4-centre pharmacophores with volumes smaller than the number of non-hydrogen atoms multiplied by this value
AREA To reject 3-centre pharmacophores with areas smaller than the number of non-hydrogen atoms multiplied by this value

Miscellaneous Changes

Saving best conformer only

By default, a site search now only saves the best conformer (based on score). If the common variable ISTALL is set to 1, and a 3D format file used (eg 3D SDfile), all conformers which score below the cutoff (GSTHIT) will be saved.

H-Donor check

In THINK v1.11, the implementation of CUSTOMISE DONOR=HYDROGEN for 3D and site searches was reversed. This meant that the value NOHYDROGEN resulted in donor atoms being checked for attached hydrogens. This has been corrected in THINK v1.12 which will result in faster searches with fewer hits (ie incorrect hits are eliminated).

3D builder

There were some circumstances when, as a result of hydrogen connections being found before carbon connections, axial ring substitutions were given preference over equatorial substitutions. The connections are now automatically re-ordered so this is no longer the case. This change will also result in some different chain conformations being generated, but will not affect the subsetc onformational search (which starts from an all extended state).

Output to “think.log”

When error messages are written to “think.log”, a date and time stamp is included.

Computation of number of conformers

The evaluation of the maximum number of conformers is important for de novo structure generation when, by default, there is a limit set in “default.lrn”. This limit eliminates molecules that are considered too flexible. To take advantage of the revised use of RCNFMX, the number of points for single and alpha bonds needs to be increased, but it is desirable that the same molecules are generated. When calculating of the maximum number of conformers within de novo structure generation and for tables, the default values of single and alpha bond points are used when the current values are above 3 and 6 respectively.

Release Notes for v1.11