A combined LS-SVM & MLR QSAR workflow for predictingthe inhibition of CXCR3 receptor by quinazolinone analogs

摘要

A novel QSAR workflow is constructed that com-bines MLR with LS-SVM classification techniques for theidentification of quinazolinone analogs as "active" or "non-active" CXCR3 antagonists. The accuracy of the LS-SVMclassification technique for the training set and test was 100%and 90%, respectively. For the "active" analogs a validatedMLR QSAR model estimates accurately their I-IP10 IC5oinhibition values. The accuracy of the QSAR model(R2 =0.80) is illustrated using various evaluation techniques,such as leave-one-out procedure (iq00 = 0.67) and vali-dation through an external test set (Rp2red = 0.78). The keyconclusion of this study is that the selected molecular descrip-tors, Highest Occupied Molecular Orbital energy (HOMO),Principal Moment of Inertia along X and Y axes PMIX andPMIZ, Polar Surface Area (PSA), Presence of triple bond(PTrplBnd), and Kier shape descriptor (1 K), demonstrate dis-criminatory and pharmacophore abilities.