Fig. 5: Ion mobility-enhanced multidimensional metabolite annotation and isomer differentiation.

a The curation of 4D metabolite library in Met4DX to support metabolite annotation. b Numbers of annotated 4D features with different confidence levels in positive and negative modes of mouse liver samples (n = 6 technical replicates). c Differentiation of isomeric metabolites with the same formula in 4D metabolite library with LC separation and LC×IM dual separations. For LC separation, ΔRT was set as ≥10 s. For IM separation, ΔCCS were set as ≥4%, ≥2%, and ≥0.5%, respectively. In our metabolite library, 125,163 of 135,638 metabolites had at least one isomeric metabolite (with the same formula), which generated a total of 16,484,071 pairs of isomeric metabolite pairs. d The pair number of co-eluted isobaric features detected in different software tools. We defined the isobaric feature pairs had Δm/z ≤ 10 ppm and ΔRT ≤ 10 s in each software tool. e The distributions of ΔCCS for co-eluted feature pairs in mouse liver samples. f–i Examples of co-eluted feature pairs with different ΔCCS detected by Met4DX but missed by MS-DIAL and MetaboScape (Supplementary Fig. 21). j Identifications of metabolite isomers in panel (f) were achieved by Met4DX. k, l Validations of N-acetyl-L-phenylalanine (k) and 3-phenylpropionylglycine (l) by chemical standards, respectively.