Advanced search
Start date
Betweenand

Clustering analysis of molecular descriptors related to octanol-water partition coefficient

Grant number: 06/00940-0
Support type:Regular Research Grants
Duration: November 01, 2006 - October 31, 2008
Field of knowledge:Physical Sciences and Mathematics - Computer Science
Principal Investigator:Waldemar Bonventi Júnior
Grantee:Waldemar Bonventi Júnior
Home Institution: Pró-Reitoria de Pós-Graduação e Pesquisa. Universidade de Sorocaba (UNISO). Sorocaba , SP, Brazil

Abstract

Lipophilicity or partition coefficient (PC) is used in Quantitative Structure-Activity Relationship (QSAR) studies and the rational molecular design as a measure of molecular hydrophobicity. This affects drug absorption, its bioavailability, hydrophobic drug-receptor interactions, molecular metabolism, as well as their toxicity. PC has become also a key parameter in environmental impact studies of chemicals. The PC can be determined by three different methods: empirically using "shake flask" (SF) method, by high pressure liquid chromatography (HPLC) or through "in silicon" methods. In this last one, predictions by specialized computer programs have been carried out through the last years based on various molecular modeling methods in drug design. The usual PC calculation methods are based on molecular fragmental constants and correction factors, reflecting intra and intermolecular interactions, whose total contribution is not justified. These methods are not converging due to the way that the molecule are fragmented and have been classified as constructionist or reductionist, giving rise to different correction factors. Since these methods are not converging and the inclusion of additional correction factors is necessary once the molecular complexity increases this project intends to deal with computational mining data techniques in order to investigate which is the molecular describers who effectively drives the values of PC. To explore relations between these molecular descriptors and PC, this project aims to explore clustering techniques to identify similarities among compounds. After that, the technique of principal component analysis (PCA) will be used to establish molecular descriptors hierarchy which can contribute for the attainment of PC driven by statistical weights. PCA will be also used to evaluate how the component variable reduction modifies the previously obtained clusters. (AU)