SwissTargetPrediction

The machine-readable formats of the cathine (C₉H₁₃NO), cathinone (C₉H₁₁NO), catheduline K2 (C₄₀H₅₁NO₁₉), and catheduline E5 (C₅₉H₆₄N₂O₂₃) structures were obtained, based on both canonical and from the PubChem Database (Kim, 2016; Kim et al., 2016). In the present study, the putative molecular targets of the cathine, cathinone, and both cathedulins were obtained using SwissTargetPrediction (http://www.swisstargetprediction.ch/) (Gfeller et al., 2014) (Supplementary Figure S1). Canonical and isomeric SMILES of cathine, cathinone, catheduline K2, and catheduline E5 were used as input sequences in the SwissTargetPrediction webserver to virtually screen the molecular targets (Daina and Zoete, 2019). The SwissTargetPrediction virtual screening tool uses the "similarity principle" to predict the most probable targets of bioactive molecules such as cathine, cathinone, catheduline K2, and catheduline E5 (Gfeller et al., 2013; Gfeller et al., 2014). In this virtual reverse screening tool, the putative binding predictions are accomplished from 376,342 experimentally active analogous compounds in 2D and 3D that strongly interact with 3,068 well-recognized protein targets (Huang et al., 2018; Daina and Zoete, 2019). In the latest version of the SwissTargetPrediction, the dataset is based on ChEMBL23, and putative protein targets are ranked based on a score that merges both 2D and 3D similarity values of an active molecule to the query molecules such as cathine, cathinone, catheduline K2, and catheduline E5 (Daina et al., 2019). Importantly, the ranking of the targets rather than the absolute values of scores or probabilities is the most meaningful parameter. A maximum of 100 probable protein targets was obtained as an output from the SwissTargetPrediction tool (Gfeller et al., 2014; Daina et al., 2019).