Standard Article

Quantitative Structure–Activity Relationships and Computational Methods in Drug Discovery

Pharmaceuticals and Drugs

  1. Alexandru T. Balaban

Published Online: 13 JUN 2008

DOI: 10.1002/9780470027318.a1918.pub2

Encyclopedia of Analytical Chemistry

Encyclopedia of Analytical Chemistry

How to Cite

Balaban, A. T. 2008. Quantitative Structure–Activity Relationships and Computational Methods in Drug Discovery. Encyclopedia of Analytical Chemistry. .

Author Information

  1. Texas A&M University, Galveston, TX, USA

Publication History

  1. Published Online: 13 JUN 2008

Abstract

Quantitative structure–activity relationships (QSARs) are mathematical equations or other types of functions (such as the weights of connections in artificial neural networks) relating chemical structures to their biological activity. The purpose of QSAR studies is to predict novel structures with either beneficial activity as drugs for human or veterinary medicine (bactericides, antiviral or anticancer drugs, and metabolic regulators, such as hypoglycemics, hypotensives, etc.), or selective toxicity for various unwanted higher organisms (pesticides such as fungicides, insecticides, acaricides, weed killers, etc.) Quantitative structure–property relationships (QSPRs) are similar, but the property may be physical or chemical. The main problems involve

  1. finding a mathematical representation of chemical structures (usually organic molecules), represented either as molecular graphs by their constitution or connectivity without considering the three-dimensional (3-D) (stereochemical) factors, or as 3-D objects including stereochemical information;

  2. measuring the biological activities of a series of molecules;

  3. finding the QSAR between each type of biological activity and the most convenient molecular descriptors.

Physicochemical parameters that are closely related to drug transport and ligand binding include lipophilicity, polarity, polarizability, and electronic influence on hydrogen-donor and hydrogen-acceptor binding. The next task is to use the QSAR to predict which novel structures to prepare in order to obtain molecules with the desired biological activity. Then the cycle is usually repeated, because a single pass seldom affords the optimal solution.

The main mathematical descriptors and the techniques for obtaining QSARs are reviewed. An important molecular parameter, which may be measured experimentally or computed from the chemical structure, is lipophilicity or hydrophobicity; Hansch introduced the n-octanol–water partition coefficient as a measure of this property, which, to a large extent, determines the ability of molecules to penetrate the lipophilic, bilayer, extra- or intracellular membranes. Lipophilicity levels that are too high lead to insolubility in water, i.e. it is difficult to administer the drug orally via the digestive tract. The main constitutional and 3-D molecular descriptors are described and examples are given. To be statistically valid, correlations must involve orthogonal or orthogonalized descriptors. Linear, multilinear, and nonlinear types of correlations are reviewed. Screening of virtual combinatorial libraries together with high-throughput combinatorial synthesis and testing provides modern tools for more efficient drug design.

Keywords:

  • QSAR;
  • isomers;
  • chemical graphs;
  • hydrophobicity;
  • molecular descriptors;
  • topological indices;
  • molecular similarity/dissimilarity;
  • computer-assisted drug design;
  • chemoinformatics