PLS Analysis for Antibacterial Activity of Natural Coumarins Using VolSurf Descriptors

In the present study the antibacterial activity of a dataset including 24 natural coumarin derivatives was investigated by means of multivariate data analysis. Antibacterial data were available against six pathogenic microorganisms. VolSurf descriptors, used to characterize structural properties, proved efficient to discriminate the compounds according to different levels of activity as shown in the PICA and PLS-DA score plots. An informative PLS model considering the six response variables in parallel with satisfactory statistics was obtained, based on 39 original descriptors (R-2 =0.86 and Q(2)=0.79). Descriptors related to the hydrophilic regions and to hydrogen bonding capability proved to exert the most significant influence on projection (VIP> 1.5) most of them contributing in a positive way to the antibacterial activity. Contrary, lipophilicity expressed as logP was found to have a negative influence. Elongation of the molecules as described by EEFR is an additional unfavorable structural feature.