Substructure Mining of GPCR Ligands Reveals Activity-Class Specific Functional Groups in an Unbiased Manner

Abstract

In this study, we conducted frequent substructure mining to identify structural features that discriminate between ligands that do bind to G protein-coupled receptors (GPCRs) and those that do not. In most cases, particular chemical representations resulted in the most significant substructures. Substructures found to be characteristic for the background control set reflected reactions that may have been used to construct this library, e.g., for the ChemBridge DIVERSet library employed these are ester and carboxamide moieties. Alkane amine substructures were identified as most important for GPCR ligands, e.g. the butylamine substructure, often linked to an aromatic system. Hierarchical analysis of targeted GPCRs revealed well-known motives and new substructural features. One example is the imidazole-like substructure common for the histamine binding receptor ligands. Another example is the planar ring system consisting of a fused five- and six-membered ring (indole-like substucture) common for the serotonin receptor ligands.

From in silico target prediction to multi-target drug design: Current databases, methods and applications

“…Frequently, protein targets are, in the spirit of chemogenomics, classified not according to sequence or fold, but according to the similarity of their ligands. Given a ligand-based classification of protein targets, one can analyze which targets are likely to be hit by a ligand, given its structure [52,53].…”

Substructure‐Based Virtual Screening for Adenosine A2A Receptor Ligands

“…These representations have been described in detail by Kazius et al [35] and van der Horst et al [36] In short, "normal" chemical representation translates chemical structures into graphs without modifications. The three "elaborate" chemical representations differ as follows: the first uses a special bond type for aromatic bonds; the second has a special atom and bond type for aromatic atoms and bonds, and the third has special types for atoms and bonds in planar ring systems.…”

“…[37] More details on the underlying algorithm are provided in references [36] and [38]. For each substructure, the number of molecules in which the substructure occurred was calculated.…”

Perspectives on Knowledge Discovery Algorithms Recently Introduced in Chemoinformatics: Rough Set Theory, Association Rule Mining, Emerging Patterns, and Formal Concept Analysis

“…van der Horst et al used it to identify substructural features important for GPCRbinding. 164 Structural fragments with skin sensitization potential were extracted using Gaston and then combined with other descriptors to construct a recursive partitioning tree for classification of lymph assay data. 165 Wang et al employed Gaston to extract SAs from carcinogenicity data.…”