Fig. 3: Structure-CCS-relationships for sulfatides.

a Comparison of experimental (gray dots: LC-TIMS-MS; black dots: MALDI-TIMS-MSI) and predicted CCS values, modeled by LipidCCS⁶¹ (red triangle), AllCCS2³⁴ (blue triangle) and DeepCCS³⁶ (green triangle). LC-MS-derived CCS values of SB1a[M-2H]^2- /SB1a[M-HSO₃]^- are marked with a star. b Strong correlation (R² = 0.9988; linear fit (red); 95% confidence interval (CI)) of LC-MS-derived and MALDI-MSI-derived ion mobility data. Mean relative deviation \(\bar{\varepsilon }\) = 0.5%. c Correlation of experimental (MALDI-MSI) and predicted (AllCCS2) CCS values. Mean relative deviation \(\bar{\varepsilon }\) is 1.4%. d Relative deviation \(\varepsilon\) reveals inconsistent deviations per subclass of the predicted CCS values using AllCCS2 against CCS values obtained by MALDI-MSI. Uncertainties were derived from for n = 4 biological replicates and expressed as standard deviation (Supplementary Table 13). e CCS values of sulfatide subclasses as a function of fatty acid (FA) chain length for three different prediction tools, AllCCS2, LipidCCS and DeepCCS yielding ambiguous relative structural relationships. f Experimental CCS values and 2^nd order polynomial fit to evaluate the contribution of the degree of glycosylation in the sulfated head group and the α-hydroxylation of the N-acyl FA. The relative positions are highlighted in the insets. g Visualization of structural relationships for the data presented in f (i), relative to ganglioside series (ii) and the contribution of saturated FA and mono-unsaturated FA. Surprisingly, the difference between SM3(O3) and SM4(O3) is reduced compared to the non-α-hydroxylated counterpart suggesting that interaction between the α-OH group and the head group may influence the three-dimensional structure. Source data is provided as a Source Data file.