Virtual Screening Using Machine Learning Approach

In this study, potential inhibitors against Harpin protein (Pectobacterium carotovorum), and Single- stranded DNA binding protein (Pseudontonas aeruginosa) is to be found. Modelled 3-D structure of target protein and their newly designed leads (inhibitors) are used for molecular docking studies using Ilex 5.1. For machine learning approach, three data sets of leads are to be formed i.e. training, dependent test and independent test and their respective physiological descriptors are identified. For virtual screening of these leads Rapid Miner 5.2.002 will be used. The support vector machine (SVM) application of this software (LibSVM), is used to make a model of training data set which will further be used to check the activity of the test data set. After this, the active leads will be considered as potential inhibitors against our target proteins. This study can thereby serve as pharmacophore for the designing of potential drugs against diseases.