Release Notes for HINT 3.06S

1. hint! 3.06S has been optimized to work with SYBYL 6.8
   hint! was compiled:
   • Under Irix 6.5 on a R10000 SGI using the -mips3 -n32 compiler directives and the Sybyl 6.8 binary shared objects.
   However, the standalone HintLogP executable (included in this release) was compiled on multiple platforms (see the release notes.)
   (See the Installation instructions for more information.)

2. This version of hint! was built from an all-new library of hint! functionality that we have been developing over the last 12-15 months. The new core algorithms are written in standard C in a toolkit structure. This enables us to much more quickly develop new hint! applications to respond to user needs, and also allows us to distribute to interested customers the means to develop their own hint! applications. See the eduSoft LC Toolkits web page for more information.

3. This is also the last major release of hint! for Sybyl that eduSoft will be distributing. We have entered into an agreement with Tripos that will transfer marketing and support for this product. A virtually identical (functionally) hint! program is slated for release by Tripos in the Sybyl 6.9.1 update. eduSoft LC customers with support contracts for hint! will be given the option to obtain a license and support from Tripos for hint!. Else, eduSoft will support this version for existing customers until support contracts expire. This is an exciting development for us, as we can now focus our energies on the science and algorithm development. Also, this will in the future allow more tight integration between Sybyl and the novel hint! capabilities. The base of hint! marketed by Tripos will continue to be our code.

4. This version includes the first implementation of HIFA (hint! Interaction Field Analysis). This is an idea that I have had for a number of years, but was actually first tried by Simon Semus (then at Wyeth-Ayerst and now at Glaxo SmithKline) and published as "A Novel Hydropathic Intermolecular Field Analysis (HIFA) for the Prediction of Ligand-Receptor Binding Affinities." S.F. Semus, Med. Chem. Res. 1999, 9, 535-550. The basic idea is that the hint! interaction fields (maps) can be used in 3D QSAR if you have the appropriate learning set -- i.e., a set of structurally characterized protein-ligand complexes. Then 3D QSAR analysis, a la CoMFA, can proceed, but instead of molecular features driving the result, it will be interaction features. So, for example, a region of space may be highlighted as unfavorable -- rectifying this would require designing compounds that would improve the ligand-biomolecule interactions in that region. While this is a considerably more demanding form of 3D QSAR, we believe it to be more specific and less subject to ambiguity of what the fields mean. As to why it took so long to implement HIFA? It became a lot easier with the toolkit structure of this version of HINT. However, it is still a bit clumsy as some method had to be devised to keep track of not only the ligands, but also their corresponding "receptors". See the new tutorial (Lesson 7) for more information.

5. Numerous other changes have been implemented as part of the code overhaul. Some small differences in results between this, and previous versions, may be observed. Following is a list of changes:
   • The "direction vectors" are now associated with the molecule's partition rather than as part of the "distance" function. Thus the choice to include directionality must be made at the molecule partition step.
   • The coded Van der Waals' radii for atoms have been somewhat revised -- this causes some changes in HINT scores when the Steric Term is active. There are now two versions of the Steric Term: the Lennard-Jones 6-9 and the Lennard-Jones 6-12. The latter is the default.
   • The smiles option has been dropped from the Sybyl-interfaced hint!. However, we are making available a standalone hint! LogP program that can interpret smiles.
   • The dynamics analysis routine and associated tutorial have been discontinued for now; however, we would like to revisit that at a future date.
   • Hint Table calculations no longer have a batch option. That was essentially irrelevant with the speed of the current generation of computers.
   • More informational warnings will be issued by this toolkit version of hint! We will summarize them at a later date. However, many have to do with a more sophisticated biomolecule interpretation that does note and report more structural inconsistencies in proteins and polynucleotides.
   • Significant improvements, especially with respect to speed, have been made in the water (rigid ligand) optimization algorithm We are slating further work in this area for upcoming releases of the hint! toolkit, which will presumably also appear in the commercial hint! releases.
   • The SPL programs, macros and expression generators have been streamlined to take advantage of the new hint executable structure.

6. We have added sugar data to the HINT partition dictionary. (We appreciate the assistance of Prof. Salvatore Guccione at Univerity of Catania, Italy.)