A data fusion approach for the rapid extraction of core scaffold information that can be used to facilitate structure determination for new psychoactive substance (NPS) tryptamines is described.
The method involves the screening of DART-HRMS data of new tryptamines against a partial least squares-discriminant analysis (PLS-DA) model that predicts the core tryptamine structure class into which the compound can be grouped. The PLS-DA prediction model was created and trained using neutral loss spectra derived from collision-induced dissociation (CID) DART mass spectral analysis of 50 tryptamine structures acquired at 60 V and 90 V, in which the sample groups were revealed by hierarchical clustering analysis (HCA). HCA of the fused neutral loss data clustered the 50 tryptamines into 10 groups based on the identities of the neutral fragments lost during fragmentation. “Leave-one-structure-out” validation of the PLS-DA model gave 100% accuracy, precision, sensitivity, and specificity. For external validation, the ability of the model to classify four compounds that were unfamiliar to it was tested, and the model was found to correctly predict the skeletal framework in each case. The results show proof of concept for how this approach can aide in the identification of new emerging psychoactive compounds. (Publisher abstract provided)