The analysis of designer drugs in human plasma is highly complex, as most of these drugs are metabolized
quickly, and often into multiple products. For novel designer drugs, it is common that reference compounds for these metabolites
are unavailable at the time of analysis. Hence, the usage of in silico procedures to accurately predict the chemical
structures of these metabolites would be very useful. In this study, the differences between several methods for prediction
of site of metabolism for cytochrome P450 mediated drug metabolism are described, and their prediction accuracies are
analyzed on a set of designer drugs. It is found that ligand-based methods, which are simpler and faster, are better than or
at least as good as much more complex structure-based methods.