Home > Technology > Protein CheminformaticsProtein Cheminformatics

Rationalize Drug Lead Optimization

Drug discovery and development is largely a slow, inefficient and costly trial and error process. Pharmaceutical companies are eagerly looking for faster, cheaper and more effective ways to accelerate the drug discovery process. Since 1970, computer aided drug design has been an integral part of the drug discover process with the introduction of computer software for quantitative structure-activity relationships (QSAR) analysis. QSAR consists of devising "descriptors" which are functions of the molecular structures of the training set (biological or chemical data) and mathematical tools to obtain a numerical model that relates the "descriptors" to biological or chemical activity. Once the mathematical relationship is generated, activities of new compounds can be predicted prior to synthesis or purchase that represents a significant conservation in both time and money. Due to several limitations, QSAR analysis, has never attained the status of being a driving force in drug discovery. APT has developed a revolutionary approach for QSAR analysis of small molecule data sets that is able to devise mathematical relationships down to atom type resolution. This approach, known as 3D-CANT (three dimensional correlation analyses) has many advantages over existing approaches. The most significant advantage of 3D-CANT is the ability to mathematically compute the contributions of each atom type in a given molecule (see figure). With this information, computational chemists can effectively determine which atom type plays the most significant role in relation to the property of the molecule. This information allows the computational scientist to rationally design a new molecule with improved properties. Unlike grid based methodologies such as CoMFA, 3D-CANT does not require super-imposable datasets, which means 3D-CANT can operate on structurally diverse molecules. Other advantages of 3D-CANT include: the ability to deal with molecules involving covalent and non-covalent interaction, molecular charges do not need to be calculated, and the models are highly predictive. In addition, a powerful and complementary software Multi-CAN, a proprietary protocol that combines the strength of structure based design and ligand based design, has also been developed for drug lead optimization.

Selected Publication

1. Cherkosov A and Chen R. (2003). 3D correlation analysis-A novel approach to the analysis of substituent effects. J. Phys Chem 107: 9695-9704.