[id] => 93
[recorddate] => 2007:01:14:16:27:12
[lastupdate] => 2007:01:14:16:27:12
[type] => article
[place] => Montreal, Quebec, Canada
[subject] => chemistry - biochemistry; chemistry - computational; informatics - applied; informatics - web programming; mathematics - modeling; mathematics - statistics
[relatedworks] =>
• 4 (some):
  • Statistical assessment of solvent mixtures models used for separation of biological active compounds, ?id=164
  • Embedded molecular geometry and molecular topology approach for structure - activity relationships, ?id=180
  • Biochemistry versus biomathematics in modelling of biological active compounds, ?id=185
[file] => ?f=93
[mime] => application/pdf
[size] => 123336
[pubname] => XXIII International Biometric Conference
[pubinfo] => International Biometric Society
[pubkey] => TP1.219 (509.pdf on CD)
[workinfo] => July 16-21
[year] => 2006
[title] => Modeling herbicidal activity of a substituted triazines class by integration of compounds complex structural information
[authors] => Ştefan ŢIGAN, Lorentz JÄNTSCHI, Sorana D. BOLBOACĂ
[abstract] =>
Relationships between herbicidal activity and structure of a set of thirty 1,3,5-substituted-triazines was studied by applying an original methodology, through integration of compounds complex structural information by the use of Molecular Descriptors Family. The obtained models (mono-, bi-, tri-, and tetra-varied models) were assessed and validated through the study of correlation coefficients, the cross validation leave-one-out scores, models stability defined as the differences between the squared correlation coefficient and the cross validation leave-one-out score, and in training versus test analysis. Comparison of the models with the previous reported model was performed by applying of the correlated correlation analysis. Analysis of the obtained models shows that the best results are obtained by the four-varied model, which has a squared correlation coefficient equal with 0.9885. The prediction abilities of the four-varied model is sustained by the cross validation leave-one-out score (0.9849), by the model stability, and by the results of training vs. test analysis (p < 0.05). Comparison of the correlation coefficient obtained by the four varied model with previous reported model shows that the four varied model obtained significantly greater value (p = 0.002). It can be conclude that herbicidal activity of 1,3,5-substituted-triazines is of geometrical and topological nature, and is strongly depended on partial change and number of directly bonded hydrogen’s.
[keywords] => Herbicidal activity; Structure-activity investigations
[acknowledgment] => Research was partly supported by UEFISCSU Romania through the project ET36/2005.