Table 1.

Dataset information used to train and test RT/CCS models

Dataset	Search	Modifications	Usage	Description
RT model
HeLa	MaxQuant	regular	training	Trypsin and LysC HeLa peptides. ref. 26
PHL		regular	testing	Pan human library. ref. 57
Phos-U2OS	Spectronaut	regular and phos	testing	Phosphopeptides of U2OS. ref. 58
CCS model
HeLa	MaxQuant	regular	training	Same as HeLa in RT section
E. coli	MaxQuant	regular	testing	E. coli peptides. ref. 26
Yeast	MaxQuant	regular	testing	Yeast peptides. ref. 26
HeLa-Open	Open-pFind	all possible PTMs	testing	Same as HeLa in RT section. Only peptides with nonregular modifications were kept after open-search for testing
Drosophila-Open	Open-pFind	all possible PTMs	testing	Drosophila peptides. ref. 26. Only peptides with nonregular modifications were kept after open-search for testing

‘regular’ in the ‘Modifications’ column refers to unmodified, Oxidation@M, Carbamidomethyl@C and Acetyl@Protein N-term. The ‘Search’ column with ‘Open-pFind’ means that we re-analyzed the MS data with Open-pFind (Methods), and only peptides with nonregular modifications were kept for testing. Otherwise, the search results were downloaded from the original publications of the datasets. RT retention time, CCS collision cross section, PTM post-translational modification.