Methods to improve ranking chemical structures in ligand-based virtual screening

Abstract

One of the main tasks in chemoinformatics is searching for active chemical compounds in screening databases. The chemical databases can contain thousands or millions of chemical structures for screening. Therefore, there is an increasing need for computational methods that can help alleviate some challenges for saving time and cost in drug discover design. The ranking of chemical compounds can be accomplished according to their chances of clinical success by the computational tools. In this paper, the techniques that have been used to improve the ranking of chemical structures in similarity searching methods have been highlighted through two categories. Firstly, the taxonomy of using machine learning techniques in ranking chemical structures have been introduced. Secondly, we have discussed the alternative chemical ranking approaches that can be used instead of classical ranking criteria to enhance the performance of similarity searching methods.