MULTIVARIATE CHARACTERIZATION AND MODELING OF THE CHEMICAL-REACTIVITY OF EPOXIDES .2. EXTENSION TO DISUBSTITUTION AND TRISUBSTITUTION

被引:7
作者
ERIKSSON, L [1 ]
VERHAAR, HJM [1 ]
SJOSTROM, M [1 ]
HERMENS, JLM [1 ]
机构
[1] UMEA UNIV,DEPT ORGAN CHEM,CHEMOMETR RES GRP,S-90187 UMEA,SWEDEN
来源
QUANTITATIVE STRUCTURE-ACTIVITY RELATIONSHIPS | 1993年 / 12卷 / 04期
关键词
PLS; MULTIVARIATE CHARACTERIZATION; CHEMICAL REACTIVITY; EPOXIDES; QSPR;
D O I
10.1002/qsar.19930120403
中图分类号
R914 [药物化学];
学科分类号
100701 ;
摘要
The chemical reactivity toward 4-nitrobenzylpyridine (4-NBP) of a series of twelve mono-, di- and trisubstituted epoxides has been modelled by means of multivariate quantitative structure-property relationships (QSPRs). The models were established using theoretical descriptor variables and partial least squares (PLS) analysis. It was found that a multivariate set of eleven descriptor variables was sufficient to adequately model the chemical reactivity. Furthermore, there existed a non-linear relationship between the theoretical descriptors and the chemical reactivity. Two-component PLS models were generally required, using the second component to compensate for non-linearity in the first. The nature of the non-linearity suggests that different physico-chemical factors are regulating the chemical reactivity among two observed subgroups of epoxides. The reactivity of one group of five epoxides with one non-substituted ring carbon is regulated by the electronegative properties of the substituent on the other ring carbon, whereas the reactivity of seven di- and trisubstituted epoxides depends on a combination of the substitution pattern around the epoxide ring and the size of these substituents. Two additional points of general relevance for QSPR modelling are also addressed, namely predictor variable selection and appropriate transformation of the dependent variable.
引用
收藏
页码:357 / 366
页数:10
相关论文
共 35 条
[1]   PREDICTIVE ABILITY OF REGRESSION-MODELS .2. SELECTION OF THE BEST PREDICTIVE PLS MODEL [J].
BARONI, M ;
CLEMENTI, S ;
CRUCIANI, G ;
COSTANTINO, G ;
RIGANELLI, D ;
OBERRAUCH, E .
JOURNAL OF CHEMOMETRICS, 1992, 6 (06) :347-356
[2]   GENERATING OPTIMAL LINEAR PLS ESTIMATIONS (GOLPE) - AN ADVANCED CHEMOMETRIC TOOL FOR HANDLING 3D-QSAR PROBLEMS [J].
BARONI, M ;
COSTANTINO, G ;
CRUCIANI, G ;
RIGANELLI, D ;
VALIGI, R ;
CLEMENTI, S .
QUANTITATIVE STRUCTURE-ACTIVITY RELATIONSHIPS, 1993, 12 (01) :9-20
[3]  
Box G.E.P., 1987, EMPIRICAL MODEL BUIL
[4]  
BOX GEP, 1978, STATISTICS EXPT
[5]  
DEBRUIJN J, 1991, THESIS U UTRECHT NET
[6]   A QUANTITATIVE STRUCTURE ACTIVITY RELATIONSHIP FOR THE ACUTE TOXICITY OF SOME EPOXY COMPOUNDS TO THE GUPPY [J].
DENEER, JW ;
SINNIGE, TL ;
SEINEN, W ;
HERMENS, JLM .
AQUATIC TOXICOLOGY, 1988, 13 (03) :195-204
[7]   THE DEVELOPMENT AND USE OF QUANTUM-MECHANICAL MOLECULAR-MODELS .76. AM1 - A NEW GENERAL-PURPOSE QUANTUM-MECHANICAL MOLECULAR-MODEL [J].
DEWAR, MJS ;
ZOEBISCH, EG ;
HEALY, EF ;
STEWART, JJP .
JOURNAL OF THE AMERICAN CHEMICAL SOCIETY, 1985, 107 (13) :3902-3909
[8]  
Dunn III W., 1984, QUANT STRUCT ACT REL, V3, P131
[9]  
ERIKSSON L, 1990, ENVIRON TOXICOL CHEM, V9, P1139
[10]  
ERIKSSON L, 1994, IN PRESS ENV TOXICOL, V13