Metabolite collision cross section prediction without energy-minimized structures

Matching experimental ion mobility-mass spectrometry data to computationally-generated collision cross section (CCS) values enables more confident metabolite identifications. Here, we show for the first time that accurately predicting CCS values with simple models for the largest library of metabolite cross sections is indeed possible, achieving a root mean square error of 7.0 angstrom(2)(median error of similar to 2%) using linear methods accesible to most researchers. A comparison on the performance of 2Dvs.3D molecular descriptors for the purposes of CCS prediction is also presented for the first time, enabling CCS prediction withouta prioriknowledge of the metabolite's energy-minimized structure.