A structural similarity networking assisted collision cross-section prediction interval filtering strategy for multi-compound identification of complex matrix by ion-mobility mass spectrometry

Ion mobility coupled with mass spectrometry (IM-MS), an emerging technology for analysis of complex matrix, has been facing challenges due to the complexities of chemical structures and original data, as well as low-efficiency and error-proneness of manual operations. In this study, we developed a structural similarity networking assisted collision cross-section prediction interval filtering (SSN-CCSPIF) strategy. We first carried out a structural similarity networking (SSN) based on Tanimoto similarities among Morgan fingerprints to classify the authentic compounds potentially existing in complex matrix. By performing automatic regressive prediction statistics on mass-to-charge ratios (m/z) and collision cross-sections (CCS) with a self-built Python software, we explored the IM-MS feature trendlines, established filtering intervals and filtered potential compounds for each SSN classification. Chemical structures of all filtered compounds were further characterized by interpreting their multidimensional IM-MS data. To evaluate the applicability of SSN-CCSPIF, we selected Ginkgo biloba extract and dripping pills. The SSN-CCSPIF subtracted more background interferences (43.24%similar to 43.92%) than other similar strategies with conventional ClassyFire criteria (10.71%similar to 12.13%) or without compound classification (35.73%similar to 36.63%). Totally, 229 compounds, including eight potential new compounds, were characterized. Among them, seven isomeric pairs were discriminated with the integration of IM-separation. Using SSN-CCSPIF, we can achieve high-efficient analysis of complex IM-MS data and comprehensive chemical profiling of complex matrix to reveal their material basis.

Automated summary

In this study, we developed a structural similarity networking assisted collision cross-section prediction interval filtering strategy for ion mobility coupled with mass spectrometry (IM-MS) analysis of complex matrix. This strategy utilizes automatic regression prediction statistics and filtering intervals to filter potential compounds and characterize their chemical structures. Experimental validation with Ginkgo biloba extract and dripping pills showed that this strategy performed better than other similar strategies, enabling efficient analysis of complex IM-MS data and comprehensive chemical profiling of complex matrix.