Combinatorial protocol in multiple linear regression/partial least-squares directed rationale for the caspase-3...

The caspase-3 inhibition activity of isoquinoline-1,3,4-trione derivatives has been analysed with the topological and molecular features from Dragon software. Analysis of the structural features in conjunction with the biological endpoints in combinatorial protocol in multiple linear regression (CP-MLR) led to the identification of 45 descriptors for modelling the activity. The study clearly suggested the role of rotatable bonds, mean information on the distance degree equality, radial centricity, bond and structural information content of five-order neighbourhood symmetry, atomic van der Waals volumes and the presence or absence of certain structural fragments to optimise the caspase-3 inhibitory activity of titled compounds. The models developed and the participating descriptors advocate that the substituent groups of the isoquinoline moiety hold scope for further modification in the optimization of the caspase-3 inhibitory activity. Analysis of these descriptors in partial least squares (PLS) highlighted their relative significance in modulating the biological response. The selected descriptors are enriched with information corresponding to the activity when compared to the remaining ones.