the Hypothetical Active Site Lattice
A Method for 3D-QSAR
Since the introduction of HASL in 1988 (J. Med. Chem. 1988, 31, 1396) this 3D-QSAR
methodology has been extensively tested and successfully applied in a wide variety of
structure-activity investigations:
- Dihydrofolate Reductase, Anti-viral Nucleoside Screening, Anti-fungal Triazines (J. Math.
Chem. 1991, 7, 273)
- Glutamine Synthase (Phosphorus & Sulfur, 1989, 45, 183)
- HIV-1 Protease (J. Med. Chem. 1994, 37, 1769)
- Anti-ulcer Imidazolyls (J. Med. Chem., 1996, in press)
- Anticoccidial Triazines, Microtubule-binding Taxols, Artemisinin Antimalarials
(unpublished results)
- DNA/Uracil Mustard Alkylation (Biochemistry, 1992, 31, 9388)
Methodology
Details of the HASL modeling paradigm can be found in recent reviews of 3D-QSAR approaches:
- Probing Bioactive Mechanisms (Magee, Henry, Block, eds) ACS Symp. Ser. 413, 1989.
- 3D-QSAR in Drug Design: Theory, Methods and Applications (Kubinyi, ed.) ESCOM, 1993.
HASL modeling is accomplished through the intermediate conversion of a molecule to a set of
regularly-spaced points (lattice) defined by Cartesian coordinates (x,y,z) and atom type. In this
way individual molecules can be compared to one another, and their lattices merged to form a
composite description called the HASL.
The activity (ED50, pKi, etc.) of each molecule is
distributed among the points in the HASL in such a way that the sum of the partial activity values
associated with a set of points belonging to each molecule is equal to the activity for that molecule.
The result is a model of the active (receptor) site consisting of points in space capable of predicting
the activities of as yet untested molecules. A recent innovation, the construction of a predictive
3D-pharmacophore, is based on the selective editing of a HASL model.
Recent investigations have
shown that HASL modeling of data sets previously analyzed using the popular 3D-QSAR
technique, CoMFA (Tripos, Inc.), has resulted in similar degrees of predictivity and insight into active
site binding. This observation is not unsurprising, since CoMFA, like HASL, relies on the
deconvolution of a complex 3D problem into equidistant points in space. While CoMFA examines
the effects (steric and electrostatic) outside the molecule, HASL focusses on a heuristic inspection
of atom types within the Van der Waal's domain of the molecule.
The new Sybyl-interfaced version of HASL has several innovative features that take unique advantage of the
power of Sybyl to create HASL models and graphically examine the model results. HASL4.0S includes an expanded
parameter set that allows two values per atom type, a new cross-validation routine, interfaces to HINT and Molconn-Z,
and new map contouring algorithms that provide a novel and informative view of the HASL model results.
More About HASL