Table 2. Preparation of data sets for training, validation and testing.

No. of data sets	Downloaded from the PDB			Local database
Substructure used	Determined		Deposited	Determined	Deposited
Pipeline used	Default	Advanced		Advanced
Model built†	2163	141	n.a.	154	n.a.
Model not built	107	40	93	5	107

^†R_free ≤ 0.45.