Forty-two molecules, thirty-eight milrinone analogues, two lead compounds, amrinone and milrinone, and two commercial products have been studied using chemometrical techniques applied to thirty theoretical descriptors and two biological activities (each one at three different concentrations). PLS Regression was applied both in the usual form PLS-I, with one response variable, and as PLS-2, with the contemporary study of more activities in the block of response variables. Regression models (both with the original activities and with log and arctan transforms) were refined by progressive elimination of conformers and of non-relevant predictors, one-at-a-time, on the basis of the relevance in the regression equation. Different sorts of model refinement gave origin to four chemometrical strategies. Special attention was deserved to the development of validation procedures for the regression analysis, in order to evaluate the true predictive ability of the refined models. The predictive optimization was based on cross-validation. Complete validation using three sets (training, optimization, external) was applied in one of the strategies. Both optimization and validation were performed in different conditions in order to eliminate the possibility of chance correlation. The severe validation procedures applied prevent underestimate of prediction error, frequently encountered when partial validation procedures are applied. Only one biological activity at the highest concentration can be predicted from the theoretical descriptors with a reasonable prediction error, measured by cross-validated explained variance. Only volume descriptors have a sure importance in the final regression model.
Chemometric study and validation strategies in the structure-activity relationships of new cardiotonic agents
BOGGIA, RAFFAELLA;FORINA, MICHELE;FOSSA, PAOLA;MOSTI, LUISA
1997-01-01
Abstract
Forty-two molecules, thirty-eight milrinone analogues, two lead compounds, amrinone and milrinone, and two commercial products have been studied using chemometrical techniques applied to thirty theoretical descriptors and two biological activities (each one at three different concentrations). PLS Regression was applied both in the usual form PLS-I, with one response variable, and as PLS-2, with the contemporary study of more activities in the block of response variables. Regression models (both with the original activities and with log and arctan transforms) were refined by progressive elimination of conformers and of non-relevant predictors, one-at-a-time, on the basis of the relevance in the regression equation. Different sorts of model refinement gave origin to four chemometrical strategies. Special attention was deserved to the development of validation procedures for the regression analysis, in order to evaluate the true predictive ability of the refined models. The predictive optimization was based on cross-validation. Complete validation using three sets (training, optimization, external) was applied in one of the strategies. Both optimization and validation were performed in different conditions in order to eliminate the possibility of chance correlation. The severe validation procedures applied prevent underestimate of prediction error, frequently encountered when partial validation procedures are applied. Only one biological activity at the highest concentration can be predicted from the theoretical descriptors with a reasonable prediction error, measured by cross-validated explained variance. Only volume descriptors have a sure importance in the final regression model.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.