Search

In isolation, a data point is only a qualified number. A collection of such qualified numbers makes a variable or descriptor. When dealing with large number of descriptors in the modeling studies for the optimum utilization of contents of the generated datasets, it is necessary to identify different models as well as Information Rich Descriptors (IRDs) corresponding to the phenomenon under investigation. Moreover, in modeling studies, each model may address different substructural regions and attributes in the predictive and diagnostic aspects of the chosen phenomenon. A study of the populationof such models provides the scope to understand the diagnostic aspects of different substructural regions and to average and extrapolate the predictive aspect beyond the individual models. Combinatorial Protocol in Multiple Linear Regression (CP-MLR) is a variable selection approach and generates multiple models to address the structure-activity relations in terms of different substructural regions and attributes in predicting the activity. The working details of CP-MLR procedure and diverse QSAR/ QSPR relations derived using this approach are reviewed. [ABSTRACT FROM AUTHOR]