Computational prediction and validation studies on a diverse dataset of cox-2 inhibitors

Abstract

In linear regression analysis, when data was derived from various reference sources, the experimental quality of such data has to be assessed. Significant variables based on the statistical data of analysis were chosen. Based on the parameters like correlation coefficient (r), F-value, cross-validation r2 etc quality of the generated equation was judged. An additional condition for high predictive ability of regression model is based on external set cross-validation r2, (R2 cv,ext) and the regression of observed activities against predicted activities and vice versa for validation set. Multivariate regression analysis using python program resulted in few influential parameters displayed significant positive and negative contribution towards biological activity of COX-2 inhibitors. A new regression model was attempted by dividing the complete set (n=64) as a 58 molecule training set and a 6 molecule validation set based on selection criteria after rejecting outliers from the data set.