How is LogP calculated by MCASE?
Our algorithm for computing LogP is based on a multiple linear regression fit to experimentally known LogP values for a large number of diverse chemical structures. The regression variables are structural descriptors, such as methyl, hydroxyl, amide, etc.. Methyl and other hydrophobic groups make a positive contribution to LogP, while hydroxyl and other hydrophilic groups make a negative contribution. LogP values for other chemicals not included in the regression data set are computed using the fitted parameters.
