| Attributs | Valeurs |
|---|
| type
| |
| Is Part Of
| |
| Subject
| |
| Title
| - GA Strategy for Variable Selection in QSAR Studies: GA-Based Region Selection forCoMFA Modeling
|
| has manifestation of work
| |
| related by
| |
| Author
| |
| Abstract
| - A novel approach using a genetic algorithm (GA) forvariable selection in comparative molecular fieldanalysis (CoMFA) was developed. This approach is named GA-basedregion selection (GARGS) since theregularly splitting regions in 3D space are used as variables insteadof each field variable. GARGS wasapplied to the data set of polychlorinated dibenzofurans (PCDF) as atest example. The number of fieldvariables was reduced from 1275 to 43, and the values ofcross-validatedr2(q2) indicatingthe internalpredictivity of the model equation was increased from 0.88 to 0.95 byGARGS. The structural requirementsfor the PCDF molecules could be easily estimated from the coefficientcontour maps of the simplifiedCoMFA model equation. These structural requirements wereconsistent with the result from the previousstudies, and the utility of GARGS was demonstrated.
|
| article type
| |
| is part of this journal
| |
