摘要:

AbstractTwo algorithms for the selection of subsets of compounds from chemical databases are presented and discussed. The first is designed to select representative subsets whilst the second is intended to select compounds which cover the available property space. Both make use of calculated physicochemical parameters in contrast to more common methods based on molecular fingerprints. This is an approach to molecular similarity which has proved successful in the past. The methods are illustrated with examples and discussed.

ISSN:0931-8771    

DOI:10.1002/qsar.19960150402

出版商:WILEY‐VCH Verlag    

年代:1996

数据来源: WILEY

摘要:

AbstractThree‐dimensional common structural features among three acaricidal compounds which inhibit respiratory complex I were investigated, assuming that they have a similar binding mode. The techniques of COMPASS (COMmon geometrical Pattern Search System) and “molecular field fitting” were employed and exhaustive conformational space searches were performed without any presumption. The results suggested that the “active conformations” of these compounds are non‐planar. Within each type a lone‐pair containing functional group on the heterocyclic ring as well as a hydrophobic region with a t‐butyl group overlapped well, despite the difference in size and chemical nature of the connecting bridge. Structure‐activity relationships of the compounds are explained successfully by the energy of the “active conformation”. The present study demonstrates the usefulness of the exhaustive conformational search to derive simultaneously the “active conformation” and th

ISSN:0931-8771    

DOI:10.1002/qsar.19960150403

出版商:WILEY‐VCH Verlag    

年代:1996

数据来源: WILEY

摘要:

AbstractThe evolutionary neural network (ENN) is a new system for modelling multifactor data. The strength of ENN's are that they can extract insignificant predictors, choose the size of the hidden layer and fine tune the parameters needed in training the network. We have used an ENN to predict the biological activities of Dihydrofolate Reductase Inhibitors. As a result, we found that evolutionary neural networks give more accurate predictions than statistical methods and feedforward neural networks.

ISSN:0931-8771    

DOI:10.1002/qsar.19960150404

出版商:WILEY‐VCH Verlag    

年代:1996

数据来源: WILEY

摘要:

AbstractEstimating the toxicity of reactive xenobiotics to aquatic organisms requires physicochemical descriptors of passive transport and chemical reactions with nucleophilic biological ligands. Herein, electrophiles whose toxic action is attributed to nucleophilic substitution (SN), Michael‐type addition and Schiff‐base formation were examined. Training sets for each molecular mechanism were generated through substructure search applied to chemicals in a fathead minnow (Pimephales promelas) database. Based on a delineation of compounds by a presumed molecular mechanism, relationships between modes of toxic action, potency (96‐hour LC50values) and mechanistically‐appropriate quantum‐chemical descriptors were explored. Monohalo‐C(sp3) function which may give rise to SNreactivity was encountered in 35 compounds. The inclusion of ELUMO, a nonspecific electrophilicity descriptor, to the generic LC50‐ hydrophobicity relation increased the explained variance from r2= 36% to 69%. Eighteen potential Michael‐type acceptors, mainly acrylates, were identified by the presence of a localized CC double bond at an α, β position to a polar group. Due to different modes of action, the toxic potency of these chemicals varies almost independently of hydrophobicity (r2= 0.12). Two additional electronic descriptors that are consistent with the likely molecular mechanism provide a multivariate QSAR with r2= 0.78. Forty‐five aldehydes and 3 formamides comprised the training set associated with probable Schiff‐base mechanism of toxicity. The results suggest a marginal increase of toxic potency from that expected due to narcosis for more electrophilic carbonyl groups. Overall, it was concluded that regressions based on data sets that combine reactive chemicals with narcotics typically require an electronic descriptor in addition to hydrophobicity, even if the compounds all contain a common electrophilic moiety related to the putative specific reaction mechanism. However, without the generation of additional toxicity data from chemical sets that incorporate a broader range of electronic and steric character, it will likely remain extremely difficult to develop a quantitative ability to predict the potency of el

ISSN:0931-8771    

DOI:10.1002/qsar.19960150405

出版商:WILEY‐VCH Verlag    

年代:1996

数据来源: WILEY

摘要:

AbstractBenzene derivatives elicit variable modes of toxic action to fish and, as a result, their toxic potency does not depend exclusively on bioaccumulation. In an attempt to develop quantitative structure‐activity relationships (QSARs) for acute toxicity across a relatively large and well‐defined chemical domain, a series of benzenes extracted from existing collections of toxicity data for guppies (Poecilia reticulata) and fathead minnows (Pimephales promelas) were studied. The selection of compounds was based on “prohibited” substructural screens that represent inherently reactive electrophilic groups or other chemical functions known to induce particular toxic effects. The resultant correlation samples were comprised of 64 and 122 substituted benzenes tested on the guppy and fathead minnow, respectively. The QSAR analysis showed that, for both samples, the toxicodynamic component of the toxic potency is mainly related to the electrophilicity of the aromatic ring in terms of the maximal acceptor superdelocalizability, Amax. By inclusion of Amaxas an additional variable into the generic log LC50‐ log Kowlinear regression, the explained variance increased from 51% to 73% and from 67% to 83% for the guppy and fathead minnow data sets, respectively. The significance of Amaxis interpreted by the fact that it provides a crude differentiation between benzenes acting through different modes of toxic action. For the guppy sample, a significant three‐variable QSAR (R2= 0.855) was derived, which employed both maximal acceptor and donor superdelocalizabilities as electronic descriptors and provided a more definite quantitative separation for the benzene

ISSN:0931-8771    

DOI:10.1002/qsar.19960150406

出版商:WILEY‐VCH Verlag    

年代:1996

数据来源: WILEY

ISSN:0931-8771    

DOI:10.1002/qsar.19960150407

出版商:WILEY‐VCH Verlag    

年代:1996

数据来源: WILEY

ISSN:0931-8771    

DOI:10.1002/qsar.19960150408

出版商:WILEY‐VCH Verlag    

年代:1996

数据来源: WILEY

ISSN:0931-8771    

DOI:10.1002/qsar.19960150401

出版商:WILEY‐VCH Verlag    

年代:1996

数据来源: WILEY