Combined Use of Factorial Design and Comparative Molecular Field Analysis (CoMFA): a Case Study
Giuseppe Caliendo, Giovanni Maria Greco, Ettore Novellino, Elisa Perissutti, Vincenzo Santagada · Quantitative Structure-Activity Relationships · 1994
Abstract The factorial design (FD) strategy has been applied within the CoMFA framework to a set of 71 N‐acyl‐L‐aminoacid esters undergoing hydrolysis by α‐chymotripsin. The steric and electrostatic fields of the molecules aligned in their enzyme‐bound conformations have been subjected to Principle Component Analysis (PCA) and three latent variables have been extracted to build up an FD scheme. Two different training sets of 12 compounds have been formed which satisfy the FD criteria. In addition to these two rationally designed series, 50 training sets of 12 structures have been defined through a random selection procedure. A CoMFA model has been derived from each training set by correlating the binding affinity to the enzyme with the ligands' steric and electrostatic fields. The efficiency of the FD approach versus a simple random selection of the structures has been finally assessed by comparing the performances of the different CoMFA models in predicting the binding affinity of 59 ligands not belonging to the training set under consideration. Our results show that in absence of a proper series design strategy the risk of deriving a poorly predictive CoMFA model cannot be neglected.