The collision cross section (CCS) values derived from ion mobility spectrometry (IMS) are a significant molecular property used for compound identification, however, accurate CCS prediction from molecular structure remains challenging. Here, the authors develop an accurate SigmaCCS approach based on graph neural networks using 3D conformers generated from SMILES strings to directly predict CCS values from molecular structures.
- Renfeng Guo
- Youjia Zhang
- Zhimin Zhang
