Optimization and Validation of a Docking-Scoring Protocol; Application to Virtual Screening for COX-2 Inhibitors

Abstract

To exploit available structural information about the cyclooxygenase enzyme for the virtual screening of large chemical libraries, a docking-scoring protocol was tuned and validated. The screening accuracy was assessed using a series of known inhibitors and a set of diverse a priori inactive compounds that was seeded with known active ligands. The major parameters of the DOCK algorithm were investigated. A large improvement of the results was obtained on tweaking some of them. The generated complexes were rescored using six scoring functions. In this way, the striking importance of this step was demonstrated, as well as the good performances of DOCK energy and SCORE for this target. The results were further improved via a consensus approach. As a first application, a subset of a large compound library was screened using this protocol. Among the compounds that were selected for biological testing, a third of them turned out to have a significant enzyme inhibition.

Fluorine in medicinal chemistry: A review of anti-cancer agents

“…More recently [106], it has been discovered that vioxx and bextra are associated with increased cases of stroke and heart diseases, and in 2004 they were withdrawn from the market. However, the fluorinated derivatives SC-558 104, and JTE-552 105, of valdecoxib and rofecoxib have been developed and are in clinical trials [107]. The therapeutic applications of selective COX-2 inhibitors in the treatment of cancers and Alzheimer disease are under investigation [108].…”

A support vector machines approach for virtual screening of active compounds of single and multiple mechanisms from large libraries at an improved hit-rate and enrichment factor

“…Machine learning (ML) methods have recently been used for developing ligand-based VS (LBVS) tools [7][8][9][10][11][12][13][14] to complement or to be combined with structure-based VS (SBVS) [1,[15][16][17][18][19][20][21][22][23][24][25][26] and other LBVS [2,[27][28][29][30] tools aimed at improving the coverage, performance and speed of VS tools.…”

Applying Computational Scoring Functions to Assess Biomolecular Interactions in Food Science: Applications to the Estrogen Receptors

“…This was proposed by Charifson [151] because the combination of diverse functions could potentially compensate for the weaknesses of individual functions, and thus yield better overall scoring. While other strategies such as post-docking minimization [93,152] or topological filtering [153] The consensus scoring model has been widely accepted [154][155][156], There are three types of consensus methods: 1) rank by "vote", where only conformations within the top x% are polled; 2) rank by rank, where the ranking from each function is averaged; and 3) rank by number, where the score value from each function is averaged.…”