Abstract

Thermolysin is a bacterial proteolytic enzyme, considered by many authors as a pharmacological and biological model of other mammalian enzymes, with similar structural characteristics, such as angiotensin converting enzyme and neutral endopeptidase. Inhibitors of these enzymes are considered therapeutic targets for common diseases, such as hypertension and heart failure. In this report, a mathematical model of Multiple Linear Regression, for ordinary least squares, and genetic algorithm, for selection of variables, are developed and implemented in QSARINS software, with appropriate parameters for its fitting. The model is extensively validated according to OECD standards, so that its robustness, stability, low correlation of descriptors and good predictive power are proven. In addition, it is found that the model fit is not the product of a random correlation. Two possible outliers are identified in the model application domain but, in a molecular docking study, they show good activity, so we decide to keep both in our database. Finally, 141 and 69 compounds (2.5 <= pK(i) < 3.5 and pK(i) >= 3.5, respectively) are identified as potential Thermolysin inhibitors, concluding that the proposed computational tools are an efficient method for the identification of new drugs that could inhibit this enzyme. (C) 2016 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University.