Home
Resource
Hybrid structure-activity relationship modeling of human cytochrome P450 isoform 2C9 inhibition
PDF
PDF format is widely accepted and good for printing.
Plug-in required
PDF-1(688.55 kb)
Citation & Export
View Usage Statistics
Staff View

Citation & Export
Hide

Simple citation

Pinolini, Daniel C.. Hybrid structure-activity relationship modeling of human cytochrome P450 isoform 2C9 inhibition. Retrieved from https://doi.org/doi:10.7282/T3Z03B3D

Export

Click here for information about Citation Management Tools at Rutgers.

Statistics
Hide

Description

TitleHybrid structure-activity relationship modeling of human cytochrome P450 isoform 2C9 inhibition
NamePinolini, Daniel C. (author); Zhu, Hao (chair); Lee, Kwangwon (internal member); Shende, Sunil (internal member); Rutgers University; Camden Graduate School
Date Created2015
Other Date2015-10 (degree)
SubjectComputational and Integrative Biology, QSAR (Biochemistry), Cytochrome P-450
Extent1 online resource (viii, 22 p. : ill.)
DescriptionHuman Cytochrome P4502C9 is a vital enzyme in human drug metabolism. Inhibition of P450 2C9 can cause critical Drug Drug Interactions (DDI). Great resources can be saved if the potential inhibition of new compounds (e.g. new drugs) can be evaluated before chemical synthesis. Computational models are promising tools to realize this goal. Previous Quantitative Structure Activity Relationship (QSAR) modeling works performed on this enzyme were not significant due to limitation of available data as training sets, and all suffer from shortcomings of traditional QSAR approaches, especially the issue of active cliffs. A successful large scale model that incorporates biological response data would be beneficial to future drug discovery. Methods In this study, QSAR modeling approaches were employed to develop multiple computational models for P450 2C9 inhibition. A training set of 20,839 compounds and an external set of 20,655 compounds were compiled from PubChem assay data. After chemical descriptors were generated for each compound, random forest and support vector machine algorithms were used to develop QSAR models based on the training set. The results of individual models were averaged as consensus predictions. Individual and consensus models were first validated using five-fold cross-validation. Then the validated models were used to predict the external sets. Results The predictivity of external set compounds for developed models was acceptable for QSAR modeling (Consensus model statistics: Sensitivity = 67.3%, Specificity = 71.3%, Correct Classification Rate = 69.3%). Incorporation of biological response data as extra descriptor information into traditional QSAR approaches improved predictivity of the associated models (Sensitivity = 67.1%, Specificity = 75.8%, Correct Classification Rate = 71.5%). These improvements were shown to be statistically significant. Conclusions In this study, QSAR models of CYP2C9 inhibition were successfully developed for a large set of compounds. Biological response data was successfully incorporated into traditional QSAR modeling procedure, leading to improvement in predictivity. This development could be used to more successfully predict the potential DDI of new compounds.
NoteM.S.
NoteIncludes bibliographical references
Noteby Daniel C. Pinolini
Genretheses, ETD graduate
Persistent URLhttps://doi.org/doi:10.7282/T3Z03B3D
Languageeng
CollectionCamden Graduate School Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.
Version 8.5.5
Rutgers University Libraries - Copyright ©2024