Comparative structural connectivity spectra analysis (CoSCoSA) models of steroid binding to the corticosteroid binding globulin

A three-dimensional quantitative spectrometric data-activity relationship (3D-QSDAR) model was developed that is built by combining NMR spectral information with structural information in a 3D-connectivity matrix. The 3D-connectivity matrix is built by displaying all possible carbon-to-carbon connections with their assigned carbon NMR chemical shifts and distances between the carbons. Selected 2D C-13-C-11 COrrelation SpectroscopY (COSY) (through-bond nearest neighbors) and selected theoretical 2D C-13-C-13 distance connectivity spectral slices from the 3D-connectivity matrix to produce a relationship among the spectral patterns for 30 steroids binding to corticosteroid binding globulin., We call this technique a comparative structural connectivity spectra analysis (CoSCoSA) modeling. A CoSCoSA principal component linear regression model based on the combination of C-13-C-13 COSY and (CC)-C-13-C-13 distance spectra principal components (PCs) had an r(2) of 0.96 and a leave-one-out (LOO) cross-validation q(2) of 0.92, A CoSCoSA parallel distributed artificial neural network (PD-ANN) model based on the combination of (CC)-C-13-C-13 COSY and (CC)-C-13-C-13 distance spectra had an r(2) of 0.96,, a leave-three-out q(3)(2) of 0.78, and a leave-ten-out q(10)(2) of 0.73. CoSCoSA modeling attempts to uniquely combine the quantum mechanics information from the NMR chemical shifts with internal molecular atom-to-atom distances into an accurate modeling technique. The CoSCoSA modeling technique has the flexibility and accuracy to outperform the cross validated variance q(2) of previously published quantitative structure-activity relationship QSAR), quantitative spectral data-activity relationship QSDAR), self-organizing map (SOM), and electrotopological state (E-state) models.