The major goal of this project is to develop novel methods of computer-assisted drug design and implement them in the context of the search for novel catechol and, especially, non-catechol ligands of dopaminergic D1 receptors. The special challenge to identify non-catechol ligands is based on the following considerations: (i) possibly reduced susceptibility to metabolic transformations characteristic of traditional dopaminergic ligands, and (ii) possible reduction of the oxidative component in dopamine neuronal depletion in current pharmacotherapy of Parkinson's disease. The success of rational drug discovery and design depends on the interpretation of experimental structure-activity relationships (SAR) data using robust and predictive analytical techniques Therefore, the proposed studies include the development and applications of novel methods of data analysis (Quantitative Structure-Activity Relationships, QSAR) and prediction (database mining, de novo design). The first part of this application concentrates on the development of novel, especially non-linear QSAR methods, and their application to known ligands of the D1 receptor. The emphasis is placed on the computational efficiency and automation of the underlying QSAR approaches. The second part emphasizes the development of efficient sampling methodologies and rational computational tools for database mining in search of chemically diverse potential ligands of dopamine D1 receptor. The last part of the application employs three-dimensional molecular models of the dopamine D1 receptor to discover new lead compounds via molecular docking and/or de novo design. The proposed methodologies employ principles of chemical similarity and diversity as grounds for similar or diverse biological action of underlying compounds. Bioactive molecules are conventionally characterized using multiple molecular descriptors, whose relative value with respect to the target property (i.e., biological activity) is unknown a priori. The proposed methodologies employ specially developed diversity/similarity functions and sampling techniques and advanced definition of pharmacophore in terms of selected descriptor variables most relevant with respect to biological activity. This descriptor-based pharmacophore is formulated on the basis of novel variable selection QSAR approaches. The long-term objective of this project is both the developments of efficient computational technologies for pharmaceutical lead generation and their application to discover novel and diverse ligands of the dopamine D1 receptor.
| Golbraikh, A; Bonchev, D; Tropsha, A (2001) Novel chirality descriptors derived from molecular topology. J Chem Inf Comput Sci 41:147-58 |
