Table 1. Numbers of SAD data sets downloaded from the PDB, excluded and used for training.

The exclusions were performed consecutively in the order of the rows (for example, if no anomalous data were present for a data set it was excluded and no other exclusion criteria were evaluated for that data set).

Total downloaded from the PDB	10891
Excluded due to no anomalous data deposited	7612
Excluded due to a different cell for the data and model	3
Excluded due to no or unusable anomalous signal	732
Total used from the PDB after exclusions	2544