In this work, Multiple Linear regression and data exploration was carried out on 16 MAP38 kinase inhibitors using the software package R. For these 16 inhibitors, we first calculated a set of five physicochemical descriptors- positive surface area, negative surface area, polar surface area, lipophilicity and weight using the MOE software. Then, explorative data analysis was carried out on this dataset, containing precalculated physicochemical descriptors using different packages available in R. Correlation analysis and Variable selection was performed to identify best subset of descriptors that influence the koff rates and it was found that 2 variables: polar surface area and weight yielded in the best linear model. Then linear regression analysis was performed to build a linear regression model and coefficients and standard errors for weight and polar surface area were determined. The final model was found to have adjusted R2 = 0.79. However due to the strong intercorrelation between weight and polar surface area, one should be careful about interpretation.
This work describes an example of using Linear Regression of Kinetic Data with Chemical Descriptors in kinetic calculations.