Documentation scienceplus.abes.fr version Bêta

À propos de : Selecting Optimally Diverse Compounds from Structure Databases: AValidation Study of Two-Dimensional and Three-Dimensional MolecularDescriptors        

AttributsValeurs
type
Is Part Of
Subject
Title
  • Selecting Optimally Diverse Compounds from Structure Databases: AValidation Study of Two-Dimensional and Three-Dimensional MolecularDescriptors
has manifestation of work
related by
Author
Abstract
  • The efficiency of the drug discovery process can besignificantly improved using designtechniques to maximize the diversity of structure databases orcombinatorial libraries. Here,several physicochemical descriptors were investigated to quantifymolecular diversity. Basedon the 2D or 3D topological similarity of molecules, the relationshipbetween physicochemicalmetrics and biological activity was studied to find valid descriptors.Several compounds wereselected using those descriptors from a database containing diversetemplates and 55 biologicalclasses. It was evaluated whether the obtained subsets representall biological properties andstructural variations of the original database. In addition,hierarchical cluster analyses wereused to group molecules from the parent database, which should havesimilar biologicalproperties. Using various sets of structurally similar molecules,it was possible to derivequantitative measures for compound similarities in relation tobiological properties. A similarityradius for 2D fingerprints and molecular steric fields was estimated;compounds within thisradius of another molecule were shown to have comparable biologicalproperties. This studydemonstrates that 2D fingerprints alone or in combination with othermetrics as the primarydescriptor allow to handle global diversity. In addition, standardatom-pair descriptors ormolecular steric fields can be used to correlate structural diversitywith biological activity.Hence, the latter two descriptors can be classified as secondarydescriptors useful for analoglibrary design, while 2D fingerprints are applicable to design ageneral library for lead discovery.Based on these findings, an optimally diverse subset containingonly 38% of the entire IC93database was generated using 2D fingerprints. Here no structure ismore similar than 0.85 toany other (Tanimoto coefficient), but all biological classes wereselected. This reduction ofredundancy led to a child database with the same physicochemicaldiversity space, whichcontains the same information as the original database.
article type
is part of this journal



Alternative Linked Data Documents: ODE     Content Formats:       RDF       ODATA       Microdata