[id] => 87
[recorddate] => 2007:01:14:15:57:19
[lastupdate] => 2007:01:14:15:57:19
[type] => conference
[place] => Varna, Bulgaria
[subject] => chemistry - computational; chemistry - organic; chemistry - physical; informatics - applied; mathematics - statistics; medicine - informatics; research - evaluation
[relatedworks] => N/A
[file] => ?f=87
[mime] => application/pdf
[size] => 313357
[pubname] => Third Humboldt Conference on Computational Chemistry
[pubinfo] => InnosLab Ltd.
[pubkey] => ISBN 954-323-199-0 & 978-954-323-199-7
[workinfo] => p. 65, June 24-28
[year] => 2006
[title] => Modeling the Octanol-Water Partition Coefficient of Substituted Phenols: the Use of Structure Information
[authors] => Lorentz JÄNTSCHI, Sorana-Daniela BOLBOACĂ
[abstract] =>
Aim: The paper presents the abilities in estimation and prediction of octanol-water partition coefficient of some para-substituted phenols through integration of complex structure information by the use of an original molecular descriptors family on structure-property relationships.
Material and Method: The measured octanol-water partition coefficient of thirty para-substituted phenols, express in logarithmical scale was taken from a previous study [T.W. Schultz, Relative toxicity of para-substituted phenols: Log KOW and pKa-dependent structure-activity relationships. Bull. Environ. Contam. Toxicol., 1987, 38, 994-9]. All para-substituted phenols were used in order to generate and calculate the molecular descriptors family [L. Jäntschi, Molecular Descriptors Family on Structure Activity Relationships 1. The review of Methodology, Leonardo Electronic Journal of Practices and Technologies, 2005, 6, 76-98]. Based on the generated descriptors, the structure-property relationships models were build. The obtained models (bi-, and four-varied) were validated through the assessment of the cross-validation leave-one-out score, and analyzed through the squared correlation coefficient and the model stability. The comparison between the bi and four-varied models was performed by applying the Steiger’s Z test.
Results: The characteristics of the bi-varied model are: the squared correlation coefficient (r2) equal with 0.8943, the leave-one-out scores (r2cv-loo) equal with 0.8659, and the model stability (r2- r2cv-loo) equal with 0.02838. The multiple linear regression analysis shows that best results are obtained in four-varied model (Ŷ= 8.69∙10-2+5.56∙10-3∙isDDkGg-0.42∙IMmrKQg+9.41∙10-3∙lPMDKQg-7.8∙10-2∙lFMMKQg): r2 = 0.9781, proving to be a stable model (r2cv-loo = 0.01) and having best prediction ability (r2cv-loo= 0.9680). The correlation coefficient obtained by the four-varied model is statistically significant higher comparing with the correlation coefficient obtained by the bi-varied model (Steiger’s Z parameter = 4.3501, p-value = 6.81∙10-6).
Conclusion: The four-varied model indicates that the octanol-water partition coefficient of studied para-substituted phenols is like to be of geometry nature, is strongly dependent on the partial change of compounds and group electronegativity and it is in relation with elastic force.
[keywords] => octanol-water partition coefficient; para-substituted phenols; structure-activity relationships