Structure annotation of small molecules by high-resolution mass spectrometry (HRMS) provides insights into the occurrence and fate of unknown organic contaminants. Various algorithms rank of molecule spectrum matches (MSMs) by predicting molecular fragmentation in tandem mass spectrometry, but such approaches generally have unknown error rates. Therefore, approaches are needed for determining false discovery rates (FDR) in MSMs. Here we describe an approach for controlling error rates in MSMs using a target-decoy strategy.  <br/><br/>Target-decoy strategies compare score distributions of candidate molecules (i.e., targets) with those derived from molecules known to be incorrect (i.e., decoys). The target-decoy strategy is common to HRMS-based proteomics, in which reversed peptide sequences comprise decoy databases. However, few approaches are available for generating small molecule decoy databases.  <br/><br/>Here, we describe an approach to generating small molecule structure decoys based on the RDKit cheminformatics software and a custom Python library (SMdecoy). Our approach defines the target database as the set of known molecules for a given molecular formula present in the PubChem database. Next, decoys are generated by mutating and rearranging molecules. Mutation generates new molecules by swapping substructures in a molecule with isomeric substructures from the set of all target and decoy molecules. Rearrangement swaps moieties within a molecule to generate new structures. After mutating and rearranging the non-redundant set of target and decoy molecules are combined and the algorithm recurses until a pre-defined number of decoys are generated. Finally, the distribution of MSM scores for target and decoy structures are evaluated based on a pre-defined FDR. In this presentation we will present evaluation of the described target-decoy strategy using MSMs for tandem mass spectra from &gt;400 unique structures using the computational MS tools MetFrag, MAGMa, and CFM-ID. Results demonstrate how FDR control can increase confidence in the high-throughput structure annotation of small molecules by HRMS.

Target-decoy strategy for controlling false discovery rates in structure annotation of small organic molecules by computational mass spectrometry

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.