Premium
Disease‐Specific Differentiation Between Drugs and Non‐Drugs Using Principal Component Analysis of Their Molecular Descriptor Space
Author(s) -
GarcíaSosa Alfonso T.,
Oja Mare,
Hetényi Csaba,
Maran Uko
Publication year - 2012
Publication title -
molecular informatics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.481
H-Index - 68
eISSN - 1868-1751
pISSN - 1868-1743
DOI - 10.1002/minf.201100094
Subject(s) - principal component analysis , drug , molecular descriptor , chemical space , disease , computational biology , profiling (computer programming) , drug discovery , pharmacology , chemistry , quantitative structure–activity relationship , computer science , bioinformatics , medicine , artificial intelligence , biology , stereochemistry , operating system
The physicochemical descriptor space has been extensively mapped and described in the literature for orally administered drugs and lead compounds. However, consideration of negative examples (non‐drugs) or disease pathophysiology is not common in many studies. In the present work, a principal component analysis was carried out using drugs and non‐drugs taking into account disease‐ and organ‐specific categories, as well as different administration routes in addition to oral. The study involves 1386 relevant small‐molecules including natural and synthetic products. Drug‐specific as well as disease‐category‐specific or organ‐specific regions and their respective threshold sets (ranges of descriptors) relative to non‐drugs were elucidated on the scores plot and validated with external, independent sets of drugs and non‐drugs. The respective loadings plot of molecular descriptors was rationalized in terms of physicochemically relevant groups related to the components of solvation free energy. The results of this analysis can contribute to the improved profiling of drug candidates and libraries making use of disease‐ and organ‐specificity coded by physicochemical descriptors and ligand binding efficiency.