摘要:

AbstractThe twenty coded amino acids have been characterized by their interaction energies, calculated by the program GRID, with six different probes mimicking various functional groups which can be involved in peptide‐peptide interactions. Principal Component Analysis (PCA) have been used to derive amino acids principal properties and comparing the amino acids classification obtained with the one derived by the previous published amino acids z‐scales. Partial Least Squares (PLS) method has been used to test the performance of GRID probe interaction energies and the newly derived principal properties in peptide's QSAR modeling. A better separation of amino acids according to the electronic features of the side chain is obtained. The inhibitory activity of the set of peptides considered has been satisfactorally model

ISSN:0931-8771    

DOI:10.1002/qsar.19930120102

出版商:WILEY‐VCH Verlag    

年代:1993

数据来源: WILEY

摘要:

AbstractAn advanced variable selection procedure, called GOLPE, aimed at obtaining PLS regression models with the highest prediction ability is presented and illustrated with an application in 3D‐QSAR. Key steps in the procedure are a preliminary variable selection by means of a D‐optimal design in the loading space, and an iterative evaluation of the effects of individual variables on the model predictivity based on the validation of a number of reduced models, on variables combinations selected according to a FFD strategy.The procedure is successfully applied to a real 3D‐QSAR case study: the results obtained by GOLPE are compared with those obtained by CoMFA and found to be in good agreement in terms of variable importance, but with a much higher prediction ability. Accordingly, the results encourage to think that it might be used within the CoMFA framework in the place of the present PLS version there, or in CoMFA‐like studies on the structures generated by GRID

ISSN:0931-8771    

DOI:10.1002/qsar.19930120103

出版商:WILEY‐VCH Verlag    

年代:1993

数据来源: WILEY

摘要:

AbstractCross validation and results with test series indicate similar predictive power for the CoMFA and the much less CPU‐time consuming MTD method. We present an MTD analysis for the binding of a series of 21 steroids to Corticosteroid Binding Globulin (CBG) and to Testosterone Binding Globulin (TeBG) series previously analyzed by CoMFA. For the construction of the hypermolecule we used crystal structures of available steroids retrieved from the Cambridge Crystallographic Data Center (CCDC) and the COSMIC software for structural manipulation and molecular superposition. The correlation results of MTD were r2= 0.93 for CBG binding and r2= 0.85 for TeBG binding. The distribution of cavity and wall vertices of the MTD receptor maps present some similarities with the CoMFA coefficient contour maps. Cross validation results of MTD (50% of data in one subseries and 50% in the second), with r2= 0.70 for CBG and r2= 0.72 for TeBG binding, are somewhat better than CoMFA results, r2= 0.66 and r2= 0.555 respectively, while CoMFA predictions for a separate series of 10 steroids are better. In CoMFA, the equation derived for the molecule receptor affinity corresponds to molecules which interact with “receptor” atoms in fixed positions by electrostatic and van der Waals 6–12 potentials. Repulsions are thereby very high for some of the van der Waals terms – while in reality the overlapping of the ligand and receptor van der Waals envelopes are circumvented by conformational transitions at lower energy costs. This may explain why the elegant CoMFA method indicates only receptor regions whose occupation by the ligand molecules have a favorable or unfavorable effect upo

ISSN:0931-8771    

DOI:10.1002/qsar.19930120104

出版商:WILEY‐VCH Verlag    

年代:1993

数据来源: WILEY

摘要:

AbstractA random source PSI electron‐accepting herbicide lead, ethyl 1,2,4‐benzotriazine‐3‐acetate1, was optimized using molecular probing, bioisosteric replacement, Free‐Wilson analysis, Sequential Simplex Optimization, and ultimately linear regression analysis. Some variables in the initial regression equations were found to be cross correlated, representing the same factors as revealed by factor analysis. The technique of prefiltering potential variables for linear regression analysis utilizing factor analysis is exemplified. For the set of1analogs, eleven independent factors were identified which were then submitted to linear regression analysis. The resulting equations are statistically more secure, all terms being independent. These equations are discussed in the context of known SAR for herbicidal PSI electron acceptors, especially the bipy

ISSN:0931-8771    

DOI:10.1002/qsar.19930120105

出版商:WILEY‐VCH Verlag    

年代:1993

数据来源: WILEY

摘要:

AbstractA multivariate QSAR for estimating the ionization potential (IP) of halo‐alkanes has been constructed. 69 compounds, including 47 for which experimental IP data in the 10 to 13.6 eV were available were parametrized with 12 structural descriptors. The data analysis to correlate IP to these descriptors was carried out by PLS that gave a tri‐dimensional model explaining over 80% of the variance of IP. The reliability of the model was evaluated on the basis of predictions by using two different approaches. These were (a) a modified PLS version providing the Standard Deviation of Errors of Prediction, the SDEP parameter, which was found to be +/− 0.45 eV, and (b) an “external” validation process where the standard deviation was computed on the basis of the predicted IP.s for 17 compounds kept outside the training set, and was found to be +/

ISSN:0931-8771    

DOI:10.1002/qsar.19930120106

出版商:WILEY‐VCH Verlag    

年代:1993

数据来源: WILEY

摘要:

AbstractQSAR models for inhibition of monoamine oxidase (MAO) by aryloxyacetohydrazide derivatives were developed for a series of MO parameters computed with the AM1 Hamiltonian. These models were compared to those based on the recently developed electrotopological state index for. skeletal atoms. The structure interpretation is the same for both types of models: the NH group adjacent to the carbonyl group together with the carbon atom, gamma to the NH group, in the hydrocarbon substituent, account for more than 90% of the computed activity. However, the model based on E‐state indexes (r2= 0.9012) is significantly superior to the one based on MO parameters (r2= 0.7985); the standard deviation for the MO based model (s = 0.27) is 50% larger than that from the E‐state based model (r = 0.19). Several points of comparison between the models are presented. This study supports the idea that the electrotopological state is a useful representation of molecular structure informat

ISSN:0931-8771    

DOI:10.1002/qsar.19930120107

出版商:WILEY‐VCH Verlag    

年代:1993

数据来源: WILEY

摘要:

AbstractProblems related to structure‐biodegradation models are discussed. They deal with the homogeneity of the data sets, the selection of an adequate statistical method, and the choice of the molecular descriptors. This short review shows that the Boolean backpropagation neural networks are promising tools to model biodegradation. To confirm this hypothesis, a heterogeneous learning set of 47 molecules and two testing sets of 23 and 17 chemicals weakly (0) or highly (1) biodegradable are described by means of 11 Boolean structural descriptors. A supervised neural network using the backpropagation algorithm generates 100% and 85% of good predictions for the learning and testing sets, respectively. The results obtained are compared with those obtained from classical regression analysis. Advantages of this new approach are given. A particular application of correspondence factor analysis is presented to transform the Boolean variables before their introduction in the neural network. Advantages of this statistical transformation are stresse

ISSN:0931-8771    

DOI:10.1002/qsar.19930120108

出版商:WILEY‐VCH Verlag    

年代:1993

数据来源: WILEY

ISSN:0931-8771    

DOI:10.1002/qsar.19930120109

出版商:WILEY‐VCH Verlag    

年代:1993

数据来源: WILEY

ISSN:0931-8771    

DOI:10.1002/qsar.19930120110

出版商:WILEY‐VCH Verlag    

年代:1993

数据来源: WILEY

ISSN:0931-8771    

DOI:10.1002/qsar.19930120111

出版商:WILEY‐VCH Verlag    

年代:1993

数据来源: WILEY