Collision cross section specificity for small molecule identification workflows
The physical-chemical property of molecular collision cross section (CCS) is increasingly used to assist in small molecule identification; however, questions remain regarding the extent of its true utility in contributing to such identifications, especially given its correlation with mass. To investigate the contribution of CCS to uniqueness within a given library, we measured its discriminatory capacity as a function of error in CCS values (from measurement or prediction), CCS variance, parent mass, mass error, and/or reference database size using a multi-directional grid search. While experimental CCS databases exist, they are currently small; thus, we used a CCS prediction tool, DarkChem, to provide theoretical CCS values for use in this study. These predicted CCS values were then modified to mirror experimental variance. By augmenting our search within a library based on mass alone with CCS at a variety of accuracies, we found that, (i) the use of multiple adducts (i.e. alternative ionized forms of the same parent compound) for the same molecule, compared to using a single adduct, greatly improves specificity and (ii) even a single CCS leads to a significant specificity boost when low CCS error (e.g. 1% composite error) can be achieved. Based on these results, we recommend using multiple adducts to build up evidence of presence, as each adduct supplies additional information per dimension. Additionally, the utility of ion mobility spectrometry when coupled with mass spectrometry should still be considered, regardless of whether CCS is considered as an identification metric, due to advantages such as increased peak resolution, sensitivity (e.g. from reducing load on the detector at any given time), improvements in data-independent MS/MS spectra acquisition, and cleaner tandem mass spectral fragmentation patterns.