Answers are grounded in this paper and cite their sources.
We describe a novel approach for the automated localization of biotransformations, which we term IsoScore. Accurate mass measurement spectra of a parent drug and its metabolites are acquired. All virtual regioisomers of a given biotransformation are generated in silico by iterating over all plausible sites of oxidation around the parent drug. Each is then fragmented virtually using an exhaustive approach supplemented with chemical intelligence. Each fragment is scored based on the likelihood that it can be formed from the precursor structure. The fragment library of each virtual isomer is then compared with the experimentally observed ions. The likelihood that a regioisomer explains the observed fragmentation data is contained in its cumulated score. We include additional weightings, which take into account the level of similarity between the mass spectra of the metabolite and the parent compound.This concept was tested on a variety of metabolites from different chemical platforms formed via single biotransformations. For a very large proportion of the metabolites, IsoScore correctly located the biotransformation to the expected position. All ions above a defined threshold in the spectrum are used to contribute to the score with no predisposition to ignore minor ions or to weight conclusions based on readily interpretable fragments. The approach is found to be most successful when differential scoring is observed between related ions in the parent and the metabolite. Further improvements in the scoring function will result in increased differentiation between likely and unlikely structures, even when the parent and the metabolite spectra show little similarity.
No takes yet. Share an insight, caveat, or question.
Leclercq et al. (2008) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: