LogP Prediction for Blocked Tripeptides with Amino Acids Descriptors (HMLP) by Multiple Linear Regression and Support Vector Regression

AbstractThe hydrophilicity/ lipophilicity of peptides are very important for rational design and drug discovery of bioactive peptides. In this study, each amino acid side chain was characterized by using three structure parameters (heuristic molecular lipophilicity potential, HMLP). Based on HMLP descriptors, prediction QSAR models of the logP were constructed for blocked tripeptides by multiple linear regression (MLR) and support vector regression (SVR). All the results showed that the logP relates to the total surface area(S) and hydrophilic indices (H), and the prediction results of SVR are better than that of MLR. The result shows HMLP parameters (S, L, H) could preferably describe the structure features of the peptides responsible for their octanol to water partition behavior