Abstract
The objective of this work was the comparison between nearest neighbor classification methods (κ-NN) and counterpropagation artificial neural networks (CP-ANN) to model the toxicity of a set of 192 organochlorine, organophosphorus, carbamate and pyrethroid pesticides, measured as Effective Concentration (EC50) and which were divided into three classes, i. e., low, intermediate and high toxicity. A total of 4885 molecular descriptors were calculated using the DRAGON program, which were simultaneously analyzed using the κ-NN method coupled with the Genetic Algorithms variable selection technique (GA-VSS). The models were appropriately validated by means of a prediction subset. The results clearly suggest that the 3D descriptors do not provide relevant information for modeling the classes. On the other hand, κ-NN shows better results than CP-ANN.
| Translated title of the contribution | Study of the Quantitative Structure-activity Relationship of Pesticides Using Classification Techniques |
|---|---|
| Original language | Spanish (Ecuador) |
| Pages (from-to) | 1-11 |
| Number of pages | 11 |
| Journal | Avances en Ciencias e Ingenierías |
| Volume | 6 |
| Issue number | 6 |
| DOIs | |
| State | Published - 19 Dec 2014 |
Keywords
- Cp-ann
- Ga-vss
- K-nn
- Pesticides
- qsar theory
CACES Knowledge Areas
- 215A Biochemistry
Fingerprint
Dive into the research topics of 'Study of the Quantitative Structure-activity Relationship of Pesticides Using Classification Techniques'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver