Identification of significant descriptors for enzyme inhibition using the LASSO method and a genetic algorithm search

Résumé fourni par la source

Enzyme inhibitors are molecules that inhibit enzyme activities. Since most drugs are enzyme inhibitors, the analysis of the properties of enzyme inhibitors is essential for effective drug discovery. In this study, we analysed the properties that are influencing inhibition in a non-redundant set of enzyme inhibitor complexes from the PDB. The binding free energies and interaction parameters of the complexes, as well as the physical and structural properties of enzymes and ligands, are found out. Then, LASSO regression using the GLMNET module of R and genetic algorithm search using BuildQSAR were performed to identify significant descriptors. Through the analysis, we identified twelve properties with a correlation of R = 0.73 and prediction accuracy was evaluated using the ROC/AUROC curve. These descriptors were tested on 255 enzyme inhibitor complexes from the MOAD database and obtained a correlation of R = 0.63. The present study finds inhibitory properties that would be useful in the initial screening of enzyme inhibition.