Application of High Performance Computing for Development of Highly Predictive 3D-QSAR Models
Abstract
Infectious diseases such as malaria, leishmaniasis and a plethora of bacterial diseases have been and continue to be among the major problems for United States Military personnel deployed in disease endemic regions of the world. We currently employ computer-aided rational drug design and discovery methods to discover new and better drugs. Here, we compute the mathematical equation correlating the observed biological activity of the drug molecule to the various descriptors, such as physicochemical properties, electrostatic and steric fields and chemical functions of the drug molecules. In brief, QSAR involves computation of the conformational model of the drug molecules, alignment of the conformers in a biologically meaningful way, computation of the descriptors, and lastly using statistical techniques such as linear regression analysis to compute the QSAR model. The traditional approach of global minimum energy conformation of the drug molecules fails to deliver good predictive QSAR models for flexible molecules. To address this issue we have developed a novel method viz. bioactive conformation mining, which consistently delivered good predictive QSAR models.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 2008
- Accession Number
- ADA505869
Entities
People
- Alan Magill
- Jayendra B. Bhonsle
- Michael Kozar
- William Mccalmont
Organizations
- Walter Reed Army Institute of Research