The Research On Surface-Based Model For Predicting Solvation Free Energy

Solvent effect plays an important role in chemical and biological processes,such as protein folding,molecular docking,as well as other interactions between bimolecular.The solvation effect is defined as the process in which a particle of solute is transferred from a fixed position in a gas phase into a fixed position in solution at constant.The key parameter to describe the effects of solvent is the free energy of solvation which is defined as the reversible work spent in the transfer process.Up to now,there have been many different solvation model,such as surface-based model,developed to address the problem that are treating solvent molecules explicitly.The surface-based model calculates solvation energy on the assumption that the overall solvation free energy is the sum of all atomic solvation contributions.Compared others,the surface based model is very charge independent and time-consuming.In the past work about surface-based model,the atom types classing based on the knowledge of chemistry was result in the serious problem of multi-correlation.A novel model named PLSOLV is proposed for predicting solvation free energy based on partial least-square regression analysis.In order to avoid the problem of multi-correlation among atomic types, the model directly extracted components from the solvent-accessible surface areas of 221 kinds atoms of amino acids by partial least squares regression.The solvation parameter for each kind of atom was obtained by fitted the predicted solvation free energy of solvation from PBSA method of 1000 proteins.PLSOLV was applied to predict the solvation free energy of 100 proteins.The predicted values were in good agreement with those from PBSA method (R=0.98).Finally,the variable importance in projection method was used to calculate VIP value of each atoms defined,and the impact on solvation free energy of the different amino acids or groups were compared.