Untitled

Abstract

SummaryThree commercially available pharmacophore generation programs, Catalyst/HipHop, DISCO and GASP, were compared on their ability to generate known pharmacophores deduced from protein-ligand complexes extracted from the Protein Data Bank. Five different protein families were included Thrombin, Cyclin Dependent Kinase 2, Dihydrofolate Reductase, HIV Reverse Transcriptase and Thermolysin. Target pharmacophores were defined through visual analysis of the data sets. The pharmacophore models produced were evaluated qualitatively through visual inspection and according to their ability to generate the target pharmacophores. Our results show that GASP and Catalyst outperformed DISCO at reproducing the five target pharmacophores.

Generation of multiple pharmacophore hypotheses using multiobjective optimisation techniques

“…In previous work, we compared the performance of GASP with Catalyst and DISCO in generating pharmacophores for sets of ligands whose binding modes have been determined experimentally, so that the true pharmacophore is 22 known [8]. This work highlighted a significant limitation of GASP that has not been addressed here, which is that GASP requires every molecule to match every feature in the pharmacophore.…”

“…This is the approach used in the commercially available programs Catalyst/HipHop [5,6] and DISCO [7]. Ensemble methods tend to produce a large number of hypotheses and require considerable effort to analyse the results [8]. Furthermore, the results are dependent on the extent to which conformational space is sampled and on the structure generation method used.…”

Chemoinformatics at the University of Sheffield 2002–2014

“…[64] GASP was developed later in collaboration with the Wellcome Research Laboratories, with the program subsequently being distributed by Certara (previously Tripos Inc.) [65] and DISCO, on a test set derived from X-ray crystal structures of protein-ligand complexes. [66] Five proteins were selected that each had crystal structures with different ligands bound. A pharmacophore was derived for each protein by superimposing the ligands according to the protein binding site.…”

“…Thus, in 1998 we were invited by the Chemical Structure Association Trust This paper has focused on research in Sheffield, but we have also been active in teaching the subject; indeed, arguably the very first textbook in the field was based on material taught in a one-semester module that Lynch developed shortly after coming to Sheffield. [81] The first full Masters programme [66] and a bibliometric study of the Journal of Chemical Information and Modeling, the core journal in chemoinformatics. [80] The lack of an up-to-date textbook led to Gillet co-authoring An Introduction to Chemoinformatics [82] (first edition was published in 2003), which became a core text for this programme as well as for the Masters courses at University of Manchester Institute of Science and Technology (now University of Manchester) and Indiana University.…”

Virtual Screening and Its Integration with Modern Drug Design Technologies

“…Machine learning methods, in which a classification rule is developed from a training set containing known active and inactive molecules. For a more comprehensive review of similarity searching, pharmacophore generation and search, machine learning and artificial intelligence approaches, the readers are referred to recent excellent reviews [22,25,[62][63][64][65].…”